Lyft, Inc. provides online ridesharing services. The Company offers ride booking, payment processing, and car transportation services. Lyft serves customers in the United States.

Lyft, Inc. operates a peer-to-peer marketplace for on-demand ridesharing in the United States and Canada. The company offers riders, personalized and on-demand access to various transportation options. It provides Ridesharing Marketplace, which enable drivers to provide their transportation services to riders. The company also offers a network of shared bikes and scooters in various cities to address the needs of riders for shorter routes; Express Drive program, a flexible car rentals program that connects drivers who need access to a car with third-party rental car companies; and concierge for organizations to manage the transportation needs of their customers and employees. In addition, it integrates third-party public transit data into the Lyft app to offer riders various transportation options; offers various enterprise programs, including monthly ride credits for daily commutes, supplementing public transit by providing rides for the first and last leg of commute trips, late-night rides home, and shuttle replacement rides; and provides transportation solutions that can be customized for events, such as recruiting events, conferences, celebrations, meetings, and company retreats. The company was formerly known as Zimride, Inc. and changed its name to Lyft, Inc. in April 2013. Lyft, Inc. was incorporated in 2007 and is headquartered in San Francisco, California.

Lyft

lyft

A rideshare co-op is launching services in Minneapolis in the coming weeks.

As Uber and Lyft threaten Minneapolis exit, driver-owned rideshare cooperative set to enter - CBS Minnesota

The multibillion-dollar rideshare companies have mobilized go-to allies to make their case to policymakers, but their lobbying spending is relatively modest.

Here's how Uber and Lyft are trying to get their way in Minnesota

Photo by on Paul Hanaoka Unsplash For the first time in five years of driving for Uber and Lyft in Minneapolis, Tom LaDuke said he’s been thinking twice before heading out. Ride-hail drivers have in a surge of carjackings throughout the city over the past couple of months. increasingly become targets Officer Garrett Parten of the Minneapolis Police Department told The Markup that the department had tracked nearly 50 carjackings of Uber and Lyft drivers since mid-August. Twelve such incidents happened in the third week of October alone. LaDuke said these numbers are unsettling. “Over the last three weeks I have not driven as much as I did before,” he said. “I’m concerned about my safety.” The incidents tend to follow a pattern, Parten said. Typically, an individual uses a stolen phone to hail a ride through Uber’s or Lyft’s app. When the driver arrives, someone meets the car and asks the driver to wait a few minutes. Shortly after, a group of “armed individuals” appears, surrounds the vehicle, and demands the driver get out. Along with the car, the group also usually takes the driver’s wallet and phone and gets the phone’s security code to . access the driver’s various accounts to transfer out money “Some drivers have been assaulted, including being pistol whipped,” read a issued by the Minneapolis Police Department on Oct. 22. “Ride-Share operators need to take extra precautions to stay safe.” citywide alert Both Uber and Lyft said they’re working to improve safety on their apps. “These reports are concerning,” said Uber spokesperson Carly DeBeikes. “Safety is a top priority for us and we are constantly working to use technology to enhance safety on the app for riders and drivers.” Lyft spokesperson Ashley Adams said, “While these horrific crimes are by no means unique to rideshare, the safety of our community is fundamental to Lyft…. We continuously invest in new features and policies to help protect drivers.” Adams and DeBeikes said the companies were working closely with the Minneapolis Police Department on the issue. But Parten said the police department had reached out to Uber and Lyft, after noting the carjacking trend, to get the companies to provide drivers with safety tips, and “they just weren’t very responsive.” The carjackings of Uber and Lyft drivers in Minneapolis fit into a . Police departments from to to have said they’ve seen an uptick of all types of carjackings during the pandemic. larger nationwide trend Chicago Cleveland Washington, D.C., In Minneapolis, carjackings and attempted carjackings increased by 339 percent over the last two years, going from 104 in 2019 to 457 so far in 2021, according to the police department. There were 401 reported carjackings in 2020. And Uber and Lyft drivers are frequent targets all over the country. In July, The Markup used police reports, local news articles, and driver interviews to . identify 124 cases of ride-hail drivers being attacked Eleven of those drivers were killed during the assaults, including Ahmed Badal, a 35-year-old Somali immigrant who was . Lyft’s Adams said the company has “attempted to contact Ahmed’s family to offer our support.” shot in April while driving for Lyft in Minneapolis Over the past few months, The Markup has identified dozens more cases of Uber and Lyft drivers being carjacked in cities including , , and . At least three more drivers have been killed. Atlanta Chicago Kansas City These numbers are far from comprehensive. Uber and Lyft have repeatedly declined to provide The Markup data on the number of carjackings involving their drivers, and most police departments don’t compile stats on carjackings that happen specifically to ride-hail drivers. Not all drivers report the incidents either. The Minneapolis Police Department’s compilation of roughly 50 carjackings since August is rare concrete evidence of the high rate at which ride-hail drivers have been targeted. Uber and Lyft treat their drivers as independent contractors rather than employees, so when carjackings occur, drivers say they’re usually . They don’t have benefits like workers’ compensation or sick leave, and losing their cars can also mean losing their livelihood. left to fend for themselves “As soon as a carjacking is reported to us, we immediately reach out to the driver to offer our support,” Lyft’s Adams said. Several drivers, however, told The Markup that after carjackings . they’ve been offered little to no help The majority of these reported carjackings happened through Uber’s and Lyft’s apps. Drivers don’t have much say in whom they pick up and where they go, and the apps make it fairly simple for passengers to to hail rides. create fake accounts with fake names In most cases, rider identification isn’t required, and things like email addresses and phone numbers can be easily spoofed. In April, Uber began requiring a valid ID in the instances when a passenger creates a new account using an untraceable payment method like a prepaid card or gift card. Lyft followed Uber’s lead in June, but only for a handful of cities, including Chicago and Atlanta. Lyft’s Adams said the company added the requirement for Minneapolis in August. Christopher Herrmann, an assistant professor at John Jay College of Criminal Justice and former crime analyst supervisor with the New York City Police Department, sees these carjackings of Uber and Lyft drivers as crimes of opportunity. “You can control the situation quite a bit,” Herrmann said, speaking from the perspective of would-be carjackers. “You’re guaranteed a nicer car. You can control the location where you get picked up. You can control what time of day it is. It’s like dial-a-carjack.” Drivers Say They’re on Their Own LaDuke first learned about Uber and Lyft drivers being carjacked in Minneapolis from a popular Twitter account called , created by volunteers who listen to police scanner audio feeds. Through those feeds, CrimeWatchMpls heard and tweeted about more than a dozen police dispatch reports of ride-hail drivers being carjacked in October. CrimeWatchMpls “Robbery of person at gunpoint. Uber driver. Loss: Vehicle, wallet phone, red Camry,” read on Oct. 4. one Twitter post Another, two weeks later, , “6:15 a.m., Lyft driver carjacked at gunpoint…. This is officially a crisis for ride share drivers in #Mpls.” read That same day, the people who run CrimeWatchMpls, who declined to comment for this story, wrote a blog post about the . “In just the last seven days (since Oct. 14), we’ve reported ten such incidents that have been aired on police dispatch. And these are just the ones we heard or know about,” read the blog post. rash of carjackings of Uber and Lyft drivers Two days later, the Minneapolis Police Department issued its citywide alert to ride-hail drivers. At that point, LaDuke said, neither Uber nor Lyft had warned Minneapolis drivers about the carjackings. On Oct. 28, nearly a week after the police alert, Uber sent drivers there a message titled “Safety tips for handling carjackings.” “Minneapolis police have reported a rise in robberies, car thefts and carjackings over the past several months,” read the message, which LaDuke shared with The Markup. The alert advises drivers to give up their vehicle and call 911 in the event of such an incident. LaDuke said he has not received any notices about the Minneapolis carjackings from Lyft. “The fact that the Minneapolis police would release this statement specific to Uber and Lyft drivers before Uber and Lyft release information had me concerned,” LaDuke said. “As a driver, we have to periodically take a selfie to show we are who we say we are, we have to have a background check. This is to keep riders safe, which is fine. But what are they doing for the drivers?” Lyft’s Adams said the company “recently sent drivers in Minneapolis emails and in-app messages highlighting safety features available to drivers and tips about staying safe on the road.” She didn’t respond to questions about whether Lyft had warned Minneapolis drivers of the carjackings specifically. Unofficial Facebook and Reddit groups where drivers in Minneapolis go to share information are filled with safety concerns. One driver in a private Facebook group said he has invested in safety gear: “1. Bullet Vest 2. Two phones 3. Dummy wallet 4. Two dash cams,” he wrote. Others say they’ve started carrying weapons, like guns and bear mace, which is prohibited by Uber and Lyft. Another driver on wrote, “I won’t pickup in North Mpls, where the majority have occurred.” Reddit According to the Minneapolis Police Department, most of the carjackings of Uber and Lyft drivers have happened in North Minneapolis, with some also occurring in Northeast and South Minneapolis. LaDuke said when he gets pinged for rides in these neighborhoods, he’s hyper-aware of his surroundings and makes sure not to stop in isolated areas or places where he could get boxed in. “It’s like I’m afraid to pick people up in that area, which is sad for me,” LaDuke said. “I grew up in North Minneapolis.” Herrmann, the assistant professor at John Jay, said most people think of their cars as a secure place and an extension of their home, so when carjackings happen, it can feel especially violating. “If someone is pointing a gun at your face and yelling at you to get out of your car, that’s a really traumatic event,” Herrmann said. “The mental scar that that leaves—every time I get in my car I think, ‘Am I going to get dragged out of my car again?’ ” Herrmann said that after carjackings a lot of people tend to change their habits, like driving less and avoiding certain areas and times of day. “The problem with gig workers is, that’s the way they make their living,” he said. “They don’t have the option or the luxury to say, ‘I’m not going to work anymore.’ ” Written by: Dara Kerr Originally published here

Many Minneapolis Uber and Lyft Drivers Were Victims of Reported Carjackings Last Fall 

Challenge: Lyft realized its was hindering its ability to deliver a consistently positive experience to its users. mobile crash reporting tool Unreliable crash reports and inefficient debugging workflows caused bugs to slip through, despite their team’s best efforts to thoroughly test and debug prior to releasing new versions of their apps. In the end, they decided to move away from Crashlytics. Results: Lyft has now standardized on Bugsnag for crash reporting across all mobile applications — iOS and Android. Bugsnag is used throughout development, testing, and post-release, allowing Lyft engineers to reliably find and resolve crashes before they can impact their users. They have even decreased the time spent troubleshooting errors by 30%. Crash Reports In a highly competitive industry, Lyft needed a trustworthy crash reporting tool to ensure the quality of application releases. Lyft is a consumer application company that facilitates peer-to-peer ridesharing, with a focus on fostering community for the drivers and passengers on their platform. For a consumer mobile application in a highly competitive industry, reliability and precision in engineering can make or break the customer experience. For this reason, the main priority for Lyft’s engineering team is reliability with the goal of “making sure new versions of our apps are fully stable and bug free so our users have a positive experience,” says Billy Pham, Technical Program Manager for Passenger Applications. However, their team lacked trust in their previous tool because, despite their well built-out processes for testing Alpha and Beta versions of their applications prior to release, bugs would still surface in major releases. They realized that missing data and error reports were causing issues in production. “We would not catch every single event, and so crashes and bugs were slipping through, and of course, this caused a lot of pain for our users. It wasn’t reliable,” says Billy. Lyft’s engineers also had difficulty assessing the impact of individual errors because there was a lack of visibility into how frequently certain errors were occurring. This was especially problematic given their fast release cycles of one week. They needed a way to understand error impact to help with decision making around release blockers. These issues, coupled with a plan to triple the size of their engineering team, led them to search for a more reliable tool that would support their hyper-growth. Observability and Prioritization for High-Quality Releases Lyft has now implemented Bugsnag across all stages of the mobile development cycle which helps them detect and fix bugs before they impact their users. Since they can deliver on their main priority of reliability, they have even made gains in productivity, shipping new releases faster than before. “Bugsnag is a great way of logging crashes reliably, efficiently assigning those issues to your teams, and resolving them in a timely manner in order to focus on new features and provide a better experience for your users,” says Billy. “Bugsnag is a great way of logging crashes reliably, efficiently assigning those issues to your teams, and resolving them in a timely manner in order to focus on new features and provide a better experience for your users.” — Billy Pham, Technical Program Manager Since implementing Bugsnag, most crashes are now caught during Alpha and Beta releases, and the time to resolution for bugs has gone down by 30%. When monitoring a release, Lyft uses filters to find errors by version, a powerful functionality that gives them visibility into each release without much effort. Dave King, Engineering Manager for Mobile Infrastructure, explains that their team relies on the quantitative data Bugsnag provides on bugs in a particular version. This visibility allows them to triage and fix all high impact bugs in a particular release without hassle. “Bugsnag has great functionality to sort and group all crashes, so we’re able to know which are the most impactful issues based on the number of events and number of users that experienced a crash,” says Billy. An Efficient Debugging Process Saves Time for Innovating Lyft’s Platform Lyft has improved their debugging process by integrating Bugsnag with JIRA and Slack, and have decreased the time spent managing bugs in their tools by 30%. Lyft uses JIRA to manage bugs and has streamlined the process by integrating with Bugsnag. The two-way sync integration allows them to avoid duplicate work because creating tickets and closing bugs across both tools happens automatically with one click. This has made prioritizing bug fixes more efficient for their team, and is one of the features they feel they could not live without. Billy explains that “this saves us a lot of time since we don’t have to copy paste and manually write descriptions. A one click button creates a ticket, and we can just pass it along to the developer. It has made a big impact on our speed.” “The Bugsnag integration saves us a lot of time since we don't have to copy paste and manually write descriptions. A one click button creates a ticket, and we can just pass it along to the developer. It has made a big impact on our speed." — Billy Pham, Technical Program Manager To improve visibility on errors throughout their team, they use Bugsnag’s API to send the top 3 errors in each platform to Slack. Receiving alerts in Slack helps the team stay informed on the status of bugs, and allows them to collaborate on fixes. It also encourages more ownership over their work since everyone is able to see incoming bugs. Another improvement to their debugging process has been adding engineers to on-call duty as they are the best placed to quickly understand and fix a bug in Bugsnag, with the added bonus of increasing ownership of their work. All these gains in efficiency taken together have in turn made it possible for their team to release product enhancements and new features 20-40% faster. Overall, Lyft “considers Bugsnag to be a better tool in terms of functionality and reliability.” Also published here.

Lyft Case Study: How to Ship New Product Features 40% Quicker 

Uber and Lyft Drivers are Increasingly Victims of Carjacking

Eva Co-op Is Continuing The Cooperative Tradition This article is based on interviews I conducted while researching for the article " Inside the Cryptocurrency Casino " in the October issue of Current Affairs. Much of the material from that article was omitted from the final article, but after several reader enquiries, the unpublished material is presented here.  I have no relationship with the Eva Coop, financial or otherwise. Joe Biden wasn’t the only one popping champagne after last year's election. While all eyes were on the presidential race, Big Tech scored a major victory with the passage of Proposition 22, which exempts drivers in the gig economy from the protection of California’s employment laws. By all accounts, it was the most expensive ballot measure in California history. Companies like Lyft, Uber, DoorDash, Postmates and Instacart spent a combined , about ten times as much as the labor groups which opposed them. $205 million Californians were “ ,” not only on television and radio but also through their smartphones. Anyone in California who opened the Uber app in the month before the election had to sit through a pro-22 ad, warning them of higher fees and longer wait times if the proposal failed. deluged with ads mandatory Afterwards, they were prompted to click “Yes on 22” in order to use the app. (There was no option to vote no, but users could click “Okay.”) As expensive as those ads were, they were a bargain compared to what the companies saved in salaries, sick leave, training, insurance and other worker protections. For most companies in the gig economy, the road to profitability comes from cutting labor costs while passing expenses like maintenance and insurance to their drivers. Uber, for example, burned $8.5 billion last year in an effort to undercut its competitors—and once they're out of business, Uber can safely raise prices, having established an effective monopoly on local markets. The system squeezes value from both drivers and riders; fewer than of Uber drivers last a full year. four percent This business model reflects the worst aspects of venture capitalism, says Dardan Isufi, co-founder and Chief Operating Officer of the Montreal-based . Eva is a ride sharing platform, similar to Uber, but with two key distinctions. Eva Co-op First, . It is owned by the riders and drivers, who vote on key decisions, elect the company’s board, and decide how to allocate revenues. Eva is structured as a cooperative Also, unlike centralized competitors, Eva’s app is based on blockchain technology, which is used to protect the data of its users and avoid centralization around a single corporate entity. Power To The Riders "We had the idea in late 2017, when Uber was bullying the government of Quebec,” Isufi explained over Zoom last Thanksgiving weekend. The government had announced legislation requiring Uber drivers to take 35 hours of online training, the same amount of training required for taxi drivers. In response, the company threatened to leave the province. Uber ultimately did not leave Quebec, but the episode provided the idea for a new kind of ridesharing platform. Along with software engineer Raphael Gaudreault, Isufi alighted on the idea of a mobility app that serves the interests of the users. “We decided to ally the cooperative principle of management to decentralization,” Isufi said. Using blockchain-based smart contracts, the app allows Eva to “give back the means of production to workers, and also to give more power to riders.” Although it only hit the road in 2019, Eva is already going places. In Montreal and Quebec City, Eva has 850 driver-members, as well as 28,000 riders, making it the second-largest mobility app in Quebec after Uber. A second Eva Coop is expected to launch in Alberta, with a third on the way in New York. These are the first outposts of what the founders hope will become a “global empire” of member-owned cooperatives. Eva Global provides the technology and initial funding for new franchises, which act as nodes on the Eva blockchain. User data is either controlled by the cooperatives, or encrypted securely in the users’ devices. The result is an app which is designed to serve the users, not investors. While Uber charges upwards of 25 percent of each fare (and UberEats takes even more) the Eva Co-op takes only fifteen percent, a figure which can be changed by a general vote of the membership. In addition, the cooperative provides free insurance to all drivers while they are operating, through a partnership with Desjardins General Insurance. And, while Eva Global acts as a parent company, it does not have the same control as Uber’s headquarters in San Francisco. Eva Global takes a maximum of five percent of total revenues, a ceiling which is encoded by smart contract and cannot be changed without agreement from the block producers. User data is decentralized: even if Eva Global were hacked, as Uber was in 2017, it would be impossible to leak any sensitive rider information. While the app is based on forked software, you don’t need to understand the blockchain works in order to hitch a ride. There are no private keys or twelve-word mnemonics: all you need is a phone and a credit card. EOS The most complicated aspects—like wallet creation, tokens and trading—are handled entirely on the back-end. “Every month I take my car and drive as a driver, just so I can chat with the customers,” Isufi says. Most riders, he says, don’t even know what a blockchain is. New Tech, Old Ideas The emphasis on cooperation and worker ownership may seem a bit out of place, considering the predominance of libertarian and capitalistic ideas in the blockchain space. Isufi, a former officer in the Canadian Army, is hard to pin down as either a wishy liberal or a doctrinaire Marxist. “I do like the capitalist system, in the sense that it allows us to innovate,” he says. “But I don’t like giving my money to Mark Zuckerberg or Jeff Bezos. They have enough money.” “Our goal is to make profit," he added. "But profit for who? For members. Not profit for some shareholders. " Canada has a long tradition of cooperative enterprise, dating back to the 1800s. Four in ten Canadians are a member of at least one cooperative organization, and the figure rises to 70 percent in Quebec. “Most Quebec citizens, their bank is a credit union,” Isufi says. “Most of our farmers are in agricultural co-ops. Some of the biggest stores in Canada are co-ops. For us, it was the ideological thing to do.” That tradition also helped Eva arrive at its initial seed funding. Instead of going to traditional venture capitalists, Eva raised its first $810,000 CAD from government bodies and cooperatives like Mouvement Desjardins, one of the largest credit unions in Canada. These investments allowed Eva’s founders to set up their offices and start hiring drivers, without giving control to financiers. But there are disadvantages to this approach. One of them is money: most investors are not motivated by social good, and driver-owned cooperatives are not exactly a high-yield investment. “Funding is our biggest weak point,” Isufi says candidly. “It’s almost a miracle that we were able to get this funding.” The founders are now targeting more traditional investors, in exchange for equity in Eva Global. There are also tradeoffs to using blockchain technology, despite efforts to keep it user-friendly. Unlike centralized platforms, Eva Co-op cannot reset your account if you get locked out. “People forget their passwords,” Isufi admitted. “And if they forget their security answer, they’re fucked.” That’s not the end of the world, since Eva Global can still create new user accounts and recover lost funds. But the process is still a turnoff for infrequent users.  “We’re losing a lot of drivers because of that,” Isufi says. “These are some of the barriers that blockchain imposes." Delivering Through The Pandemic The COVID-19 pandemic threw another wrench in the works, shuttering businesses and halting tourism.  When I interviewed Dardan Isufi in October, Quebec was reporting a thousand cases per day, and every restaurant and bar in Montreal was closed. Even houseguests were forbidden. But the health crisis also presented an opportunity to explore a new avenue of business: instant deliveries. “What we did is create a software that can integrate any online website,” Isufi explained, which can estimate delivery times and fees. “Even Amazon Prime cannot do better than one business day, but we can do like thirty-four minutes.” There are no credit card fees or account creation: after placing an order, customers can track deliveries in real time, call the driver, and even change the destination. Moreover, unlike the 30 percent commission by UberEats, Eva Co-op only charges merchants a flat monthly fee. “Since we’ve launched that, forty percent of our market share is instant deliveries,” Isufi says, a figure which he estimates at about 2,000 deliveries per month, for 300 merchant clients.  “I think the money-maker of Eva will be instant delivery.” Can Co-ops Beat Uber? Despite the pandemic, the Eva Coop in Quebec is continuing to onboard new riders. Ironically, Isufi says, Uber has turned into one of their best marketing tools: many drivers work on both apps, and introduce their other riders to the local ride sharing co-op. The developers are also working on new functionality for the application’s users. While cryptocurrency payments are a distant goal, the priority right now is to introduce features like tipping and shared invoices, which may appeal to a mass market. While the technology is unique, Eva’s founders see the project as more than just an interesting application of blockchain technology: it’s also a strong counterpoint to the Amazons and Ubers of the tech world, which derive value at the expense of their workers. If Eva is successful, it will be a powerful demonstration that an enterprise can be both profit-oriented and socially responsible. “Every company should be community oriented,” Isufi says. “Why should we empower some already-rich people?”

Uber, Lyft, & Co. Spent $205M On Proposition 22: Here's What Men Can Do Against Such Reckless Hate

Sometime ago for a project, I was looking for a good rate-limiting service. For the scope of that project, the service would run along a front proxy and would rate-limit requests to third party applications. Nginx Plus and Kong certainly have rate-limiting features but are not OSS; while I am a bigger fan of OSS. Using Istio service mesh would have been a overkill. Therefore I decided to use . Envoy Proxy + Lyft Ratelimiting The aim for this blog is help you get started with the rate-limiting service and configure various combinations of rate-limiting scenarios. Let’s dive in… Understanding Lyft Ratelimiter Ratelimit configuration consists of : A domain is a container for a set of rate limits. All domains known to the Ratelimit service must be globally unique. They serve as a way for different teams/projects to have rate limit configurations that don’t conflict. Domain : A descriptor is a list of key/value pairs owned by a domain that the Ratelimit service uses to select the correct rate limit. Descriptors are case-sensitive. Examples of descriptors are: Descriptor (“database”, “users”) (“message_type”, “marketing”),(“to_number”,”2061234567") (“to_cluster”, “service_a”) (“to_cluster”, “service_a”),(“from_cluster”, “service_b”) Descriptors can also be nested to achieve more complex rate-limiting scenarios. We will be performing rate limiting based on various HTTP headers. Let’s have a look at the configuration file. domain: apis
descriptors:
  - key: generic_key
    value: global
    rate_limit:
      unit: second
      requests_per_unit: 60
  - key: generic_key
    value: rate_limit:
      unit: second
      requests_per_unit: 50
  - key: header_match
    value: rate_limit:
      unit: second
      requests_per_unit: 40
  - key: header_match
    value: rate_limit:
      unit: second
      requests_per_unit: 30
  - key: header_match
    value: post
    rate_limit:
      unit: second 
      requests_per_unit: 20
  - key: header_match
    value: get
    rate_limit:
      unit: second 
      requests_per_unit: 10
  - key: header_match
    value: path
    rate_limit:
      unit: second 
      requests_per_unit: 5 - key: custom_header
    descriptors:
    - key: plan
      value: BASIC
      rate_limit:
        requests_per_unit: 2
        unit: second
    - key: plan
      value: PLUS
      rate_limit:
        requests_per_unit: 3
        unit: second local "123" "456" #Using nested descriptors In the configuration above it can be clearly seen There are various different keys with different ratelimit values. We can use them globally for the entire vhost in envoy or even locally for a particular path. We can also have nested values of descriptors. Envoy Front proxy Let’s see how we can use them in the envoy config now. Have a look at the configuration below: static_resources:
  listeners:
  - name: listener_0
    address:
      socket_address:        
        address: 0.0.0.0
        port_value: 10000
    filter_chains:
    - filters:
      - name: envoy.http_connection_manager
        config:
          stat_prefix: ingress_http
          codec_type: AUTO
          route_config:
            name: local_route
            virtual_hosts:
            - name: nginx
              domains:
              - rate_limits:
                - stage: 0
                  actions:
                    - generic_key: 
                        descriptor_value: routes:
              - match:
                  prefix: route:
                  cluster: nginx_1
                  include_vh_rate_limits: rate_limits:
                    - actions:
                        - generic_key:
                            descriptor_value: - actions:
                        - header_value_match:
                            descriptor_value: headers: 
                              - name: prefix_match: - actions: - requestHeaders:
                          descriptor_key: header_name: - requestHeaders:
                          descriptor_key: header_name: - match:
                  prefix: route:
                  cluster: nginx_2
                  include_vh_rate_limits: rate_limits:  
                    - actions:
                        - generic_key:
                            descriptor_value: - actions:
                        - header_value_match:
                            descriptor_value: headers: 
                              - name: prefix_match: - actions:
                        - header_value_match:
                            descriptor_value: headers: 
                              - name: prefix_match: - actions:
                        - header_value_match:
                            descriptor_value: headers: 
                              - name: prefix_match: - actions:
                        - header_value_match:
                            descriptor_value: headers: 
                              - name: prefix_match: http_filters:
          - name: envoy.rate_limit
            config:
              domain: apis
              failure_mode_deny: rate_limit_service:
                grpc_service:
                  envoy_grpc:
                    cluster_name: rate_limit_cluster
                  timeout: 0.25s

          - name: envoy.router

  clusters:
  - name: nginx_1
    connect_timeout: 1s : strict_dns
    lb_policy: round_robin
    hosts:
    - socket_address:
        address: nginx1
        port_value: 80

  - name: nginx_2
    connect_timeout: 1s : strict_dns
    lb_policy: round_robin
    hosts:
    - socket_address:
        address: nginx2
        port_value: 80

  - name: rate_limit_cluster : strict_dns
    connect_timeout: 0.25s
    lb_policy: round_robin
    http2_protocol_options: {}
    hosts:
    - socket_address:
        address: ratelimit
        port_value: 8081

admin:
  access_log_path: address:
    socket_address:
      address: 0.0.0.0
      port_value: 9901 "*" "global" "/nginx_1" true "local" "get" ":method" "GET" #This will be triggered if `X-CustomHeader` is present AND the X-CustomPlan header has a value of either BASIC or PLUS "custom_header" "X-CustomHeader" "plan" "X-CustomPlan" "/nginx_2" true "local" "123" "X-MyHeader" "123" "456" "X-MyHeader" "456" "post" ":method" "POST" "path" ":path" "/nginx" false type type type "/dev/null" Here is how it works: We have defined a named which matches all domains. single vhost nginx There is global rate-limit defined for this vhost. The . descriptor value is global Next, we have under this vhost. Namely, . Routes for cluster and similarly for nginx2. 2 clusters nginx1 and nginx2 path /nginx1 are routed to nginx1 For nginx1, there is generic rate limit defined by descriptor value local and then we have rate-limits for different values of . standard HTTP headers such as method, path etc., and some custom HTTP headers such as X-CustomHeader We have similar rate-limiting set for nginx2 cluster. These 2 nginx clusters defined here actually refer to 2 different nginx containers running as part of docker-compose stack. The overall architecture can be visualized as Simulation Architecture Since they are running on the same network as the ratelimiter and envoy proxy, they can be accessed easily using the container name. All configuration files for this set-up can be found here . In order to run the set-up just clone the repo and do docker-compose up Once the stack is up, you will have 2 nginx containers running on port 9090 and 9091 of localhost.Envoy proxy to intercept and relay requests to nginx servers. Envoy admin console can be reached at localhost:9901. Envoy will be listening as localhost:10000. Ratelimiting service container with configured rate limits which will be used to envoy. Redis container which is used by the Ratelimiting service. It is important to understand that all the applicable actions for a particular path in a cluster are aggregated by the ratelimiter for the result i.e., Logical OR of all the applicable limits Let’s Test our Setup Firstly we need to install , a load testing framework. It can be done by Vegeta brew update && brew install vegeta : request on at 100 requests per second : 10% requests successful. (Logical of , and ) : Test Scenario 1: Case GET /nginx_1/ Expected Result OR "descriptor_value": "global" "descriptor_value": "local" "descriptor_value": "get" Command echo "GET http://localhost:10000/nginx_1/" | vegeta attack -rate=100 -duration=0 | vegeta report Actual Result $ | vegeta attack -rate=100 -duration=0 | vegeta report
          Requests      [total, rate, throughput]  1008, 100.12, 10.92
          Duration      [total, attack, ]      10.071526192s, 10.06832056s, 3.205632ms
          Latencies     [mean, 50, 95, 99, max]    4.718253ms, 4.514212ms, 7.426103ms, 9.089064ms, 17.916071ms
          Bytes In      [total, mean]              68640, 68.10
          Bytes Out     [total, mean]              0, 0.00
          Success       [ratio]                    10.91%
          Status Codes  [code:count]               200:110  429:898  
          Error Set:                               429 Too Many Requests echo "GET http://localhost:10000/nginx_1/" wait — — — : : request on at 100 requests per second. : 50% requests successful. (Logical of and ) : : Test Scenario 2 Case POST /nginx_1/ ExpectedResult OR "descriptor_value": "global" "descriptor_value": "local" Command echo "POST http://localhost:10000/nginx_1/" | vegeta attack -rate=100 -duration=0 | vegeta report ActualResult $ “POST http://localhost:10000/nginx_1/ echo " | vegeta attack -rate=100 -duration=0 | vegeta report
 Requests [total, rate, throughput] 4344, 100.02, 50.56
 Duration [total, attack, wait] 43.434227783s, 43.429286664s, 4.941119ms
 Latencies [mean, 50, 95, 99, max] 5.190485ms, 5.224978ms, 7.862512ms, 10.340628ms, 20.573212ms
 Bytes In [total, mean] 1370304, 315.45
 Bytes Out [total, mean] 0, 0.00
 Success [ratio]  50.55%
 Status Codes [code:count] 200:2196 429:2148 
 Error Set:                429 Too Many Requests — — — Case: request on at 100 requests per second with : 5% requests successful (Logical of , , , and ) : : Test Scenario 3: GET /nginx_2/ X-MyHeader: 123 Expected Result OR "descriptor_value": "global" "descriptor_value": "local" "descriptor_value": "123" "descriptor_value": "path" Command echo "GET http://localhost:10000/nginx_2/" | vegeta attack -rate=100 -duration=0 -header "X-MyHeader: 123" | vegeta report Actual Result $ | vegeta attack -rate=100 -duration=0 -header | vegeta report
          Requests      [total, rate, throughput]  3861, 100.03, 5.18
          Duration      [total, attack, ]      38.604406747s, 38.597776398s, 6.630349ms
          Latencies     [mean, 50, 95, 99, max]    4.96685ms, 4.673049ms, 7.683458ms, 9.713522ms, 16.875025ms
          Bytes In      [total, mean]              124800, 32.32
          Bytes Out     [total, mean]              0, 0.00
          Success       [ratio]                    5.18%
          Status Codes  [code:count]               200:200  429:3661  
          Error Set:                               429 Too Many Requests echo "GET http://localhost:10000/nginx_2/" "X-MyHeader: 123" wait — — — : : request on at 100 requests per second with : 5% requests successful (Logical of , , , , and ) : : Test Scenario 4 Case POST /nginx_2/ X-MyHeader: 456 Expected Result OR "descriptor_value": "global" "descriptor_value": "local" "descriptor_value": "post" "descriptor_value": "456" "descriptor_value": "path" Command echo "POST http://localhost:10000/nginx_2/" | vegeta attack -rate=100 -duration=0 -header "X-MyHeader: 456" | vegeta report Actual Result $ “POST http://localhost:10000/nginx_2/ echo " | vegeta attack -rate=100 -duration=0 -header “X-MyHeader: 456” | vegeta report
 Requests [total, rate, throughput] 2435, 100.04, 5.13
 Duration [total, attack, wait] 24.346703709s, 24.339554311s, 7.149398ms
 Latencies [mean, 50, 95, 99, max] 5.513994ms, 5.255698ms, 8.239173ms, 10.390515ms, 20.287931ms
 Bytes In [total, mean] 78000, 32.03
 Bytes Out [total, mean] 0, 0.00
 Success [ratio] 5.13%
 Status Codes [code:count] 200:125 429:2310 
 Error Set: 429 Too Many Requests — — — : : request on at 100 requests per second with and : 3% requests successful (Logical of , , , , and ) : : Test Scenario 5 Case GET /nginx_1/ X-CustomHeader: XYZ X-CustomPlan: PLUS Expected Result OR "descriptor_value": "global" "descriptor_value": "local" "descriptor_value": "get" "descriptor_key": "custom_header" "descriptor_key": "plan" Command echo "GET http://localhost:10000/nginx_1/" | vegeta attack -rate=100 -duration=0 -header "X-CustomHeader: XYZ" -header "X-CustomPlan: PLUS" | vegeta report Actual Result $ “GET http://localhost:10000/nginx_1/ echo " | vegeta attack -rate=100 -duration=0 -header “X-CustomHeader: XYZ” -header “X-CustomPlan: PLUS” | vegeta report 
 Requests [total, rate, throughput] 2372, 100.05, 3.16
 Duration [total, attack, wait] 23.71156424s, 23.707710415s, 3.853825ms
 Latencies [mean, 50, 95, 99, max] 5.396743ms, 5.179951ms, 7.981171ms, 10.084753ms, 15.086778ms
 Bytes In [total, mean] 46800, 19.73
 Bytes Out [total, mean] 0, 0.00
 Success [ratio] 3.16%
 Status Codes [code:count] 200:75 429:2297 
 Error Set: 429 Too Many Requests — — — : : request on at 100 requests per second with and : 10% requests successful (Logical of , , ) : : Test Scenario 6 Case GET /nginx_1/ X-Header: XYZ X-CustomPlan: PLUS Expected Result OR "descriptor_value": "global" "descriptor_value": "local" "descriptor_value": "get" Command echo "GET http://localhost:10000/nginx_1/" | vegeta attack -rate=100 -duration=0 -header "X-Header: XYZ" -header "X-CustomPlan: PLUS" | vegeta report Actual Result $ “GET http://localhost:10000/nginx_1/ echo " | vegeta attack -rate=100 -duration=0 -header “X-Header: XYZ” -header “X-CustomPlan: PLUS” | vegeta report
 Requests [total, rate, throughput] 1578, 100.07, 10.78
 Duration [total, attack, wait] 15.773500478s, 15.769158977s, 4.341501ms
 Latencies [mean, 50, 95, 99, max] 4.748516ms, 4.514902ms, 7.076671ms, 8.518779ms, 16.077828ms
 Bytes In [total, mean] 106080, 67.22
 Bytes Out [total, mean] 0, 0.00
 Success [ratio] 10.77%
 Status Codes [code:count] 200:170 429:1408 
 Error Set: 429 Too Many Requests Scalability In a production scenario, you might want to run multiple instances of your proxy which can refer to the same ratelimit cluster. The front proxy is basically stateless. As far as the ratelimiting service is concerned, I would recommend scaling it horizontally and moving out Redis Cache to a cloud-based service like RedisLabs or AWS Elastic-cache. Also, using separate is highly recommended. All you need to do is: Redis for Per Second Limits Set the env var, : REDIS_PERSECOND "true" Set Redis endpoint, REDIS_PERSECOND_URL Conclusion We can clearly see that the actual results are pretty close to expected results. We can accomplish all kinds of complex rate limiting scenarios using this and perform request throttling. Feel free to reach out should you have any questions around this. This article was also published on . https://appfleet.com/blog/http-throttling-using-lyft-global-ratelimiting/

HTTP Throttling via Lyft Global Rate Limiting

Let us all admit it! We have always been thrilled by rivalries between two fitting competitors. We have seen it happen between Lionel Messi and Cristiano Ronaldo. We have seen it happen between Roger Federer and Rafael Nadal. The world of technology and business cannot be an exception to this pattern. When it comes to mobile devices and operating systems, it is Google versus Apple. When it comes to enterprise computing, Apple has always been at loggerheads with Microsoft. However, all these technological rivals belong to the old school if we take into consideration the revolution called on demand transportation. When it comes to on-demand transport, especially in the United States, the competition has always been between Uber and Lyft. Any regular person in the United States who has been used to availing on-demand taxi services surely have heard these names being mentioned parallel to each other. This is quite possible that they might even have both the mobile apps installed on their phone. However, they are different as much as they are similar. Both of them have their own share of pros and cons. There are also differences when it comes to their approach and business model. In this write up, let us look at the key differences between Uber and Lyft. Even though they provide the same kind of services they differ in various aspects. The crust of it might seem to be identical. The core of it is not. Where it all starts… Uber, as an idea, originated in France in 2008. However, it started as a company in San Francisco a year later. Lyft was a bit late to the party and it started only in 2012. However, the delay has not deterred Lyft from gaining a considerable share in the US market. The business model The differences between Uber and Lyft is quite a good example that can be given for how businesses that offer almost the same services can have a drastically different set of business models. Uber has always been towards the concept of professional driving services. Most of their audience are business-centric. Uber has also been looking to expand its offerings into multiple segments not just within transportation of people but even in tracking and delivery of food. Lyft, on the other hand, focuses on creating a friendly relationship between the driver and the customer. It is quite likely that the intensity of engagement is far ahead in Lyft compared to an Uber. In addition to that, Lyft has not scattered its focus anywhere beyond cab services. Explicitly, their management has stated that the aim of Lyft is to be the best when it comes to transportation services. In terms of revenue Even when it comes to revenue, there is quite a massive difference. Uber has been dominating the market and it is only expected that Uber has a higher revenue. However, when it comes to the take rate Lyft performs far ahead of Uber. While the take rate of Uber has been struggling to cross the 20% mark, Lyft has been contemplating on breaching the 30% mark. Their application Both these businesses have dedicated apps for Android, iOS and even Windows. There are a lot of similarities between these apps as they offer the same service. Apart from functionality, both these apps offer an amazing user experience. Uber is rich in features and some of these features make it a better option than Lyft. For example, in Uber, the fair estimates and the ETA details are provided even before the ride starts making it extremely transparent. When it comes to Lyft, these details are not provided. Since Lyft is light on features, it becomes extremely easy for making the user interface intuitive and easy to navigate, making it a clear winner in these aspects. The magnitude of engagement When we talk about magnitude, it is best that we measure it statistically! Based on US Android data provided by reliable sources, Uber has an open rate less than Lyft. Lyft has an open rate of 4.5% but Uber stagnated at 2.3%. The best or the worst part (depending on whether or not you are on the side of Uber) is that the numbers are expected to tilt Lyft. What makes Lyft win this game is its intense referral program. The referral program not only targets customers but also drivers. In fact, Lyft offers better referral incentives and sign up rewards, making it a lucrative option for drivers to refer and get referred. This has been one of the greatest contributing factors when it comes to the quality of services and the magnitude of engagement. The vehicle options Both Uber and Lyft give quite a wide range of vehicle options to choose from. Uber offers six different options whereas Lyft offers four. What makes Uber special in this regard is that they also have vehicles that offer a wheelchair accessibility, making it quite accommodative for the differently abled. Lyft counts only on the utility but not on the fancy names. The vehicle options of Uber might seem great on the surface but what Lyft has managed to master is in the art of decluttering. People, more often than not, do not want to experience different levels of luxury but Uber has created a few needless layers of luxury that could have been done away with. Funding and market cap By the time we started to talk about this, we are sure you would have guessed that there is no need for a comparison here. With the valuation of Uber crossing $60 billion and the valuation of Lyft struggling to beat $15 billion, Uber is quite the clear winner. However, when it comes to the funding and the revenue, Lyft has kept it a bit low profile. Uber has raised more than $24 billion in 22 rounds of funding whereas Lyft has raised about $5 billion in 19 rounds. What is to be noted here is that the growth rate of Lyft is far ahead of Uber. Lyft has managed to double its market cap within a year while steadily gaining on the market share. Market share We have now come to discuss the final aspect of our differences which is in terms of the market share. It is quite evident that Uber dominates the market in the United States. It goes without saying that, with the existing marketing effort and global recognition, Uber dominates the market. However, in the past few years, Lyft has been steadily catching up while Uber has been losing out. In 2017, Uber had a 74% market share but it had dropped to 69% the very next year. At the same time when  Lyft started with a market share of 22% but started to grow steadily, reaching a whopping 29%. Conclusion Uber and Lyft have been at constant yet healthy loggerheads with each other. It is only the presence of a competitor like Uber that makes Lyft change their game and give services in a better way. On the other hand, it is the presence of competitors like Lyft that demanded Uber to go for competitive rates and addition of better features. As an aspiring entrepreneur who would like to create a taxi business out of the app, you should ensure that you take the best of both worlds and incorporate them in your app, so your business finds immediate success! Uber clone

Uber Cool vs. Face Lyft

Last year, Uber announced that it will start showing Lime scooters and Jump bikes in Uber app. On surface, it looks like this move will cannibalise Uber’s cab business. Per trip revenue made by Uber is much higher in case of a cab than in case of an e-scooter or a bike. https://twitter.com/uber/status/1016348927312965633?lang=en Six months later, Uber went one step further and started showing public transit information on the app. Now users can get to their preferred destination via Uber using public transport, which is non-monetised by Uber. https://twitter.com/uber/status/1091022619028176896?lang=en A week ago, Lyft followed suit and announced a similar experience on its platform with an arguably better user experience (bottom bar). https://twitter.com/lyft/status/1176650044293824512?lang=en So, why are Uber and Lyft adding commute options on their app that are potential competitors and will cannibalize their core business — cabs? To answer this we will have to take a step back and look at what is the “job” these apps are trying to accomplish for its users. Theodore Levitt, the former Harvard Business School professor, famously said: “People don’t want to buy a quarter inch drill. They want to buy a quarter inch hole”. The premise of jobs to be done theory is that effective customer segmentation doesn’t happen by looking at customer attributes or product features, but by getting a deeper understanding of what job user is trying to accomplish. People have “jobs” that arise in their day to day life, that need to be taken care of. When people come across a job that needs to be done, they look for a product they can “hire” to get the job done. Users don’t buy a product to get product features. They hire a product to get a specific job done, in a specific circumstance. technique, made popular by Toyota, is very helpful while analysing jobs to done. The gist of 5 whys technique is that you'll find the root cause of a problem by asking multiple times. The 5 Whys why? Imagine a situation in which a person want to buy a quarter inch drill. The 5 why approach would be to ask the person why does he want to buy a drill. He might get back with a reason like: to make a hole in the wall. — Why do you want to make a hole in the wall? — To hang a picture of my family on the wall. Q A — Why do you want to hang a picture of your family on the wall? — To give a personal/emotional feel to the house. In this circumstance, the job that a quarter inch drill is trying to do is to make the home personal. Q A The competitor of quarter inch drill in this circumstance is a sticky hinge that stays on a wall with a glue. Another potential competitor for a drill in this circumstance can be a digital picture frame, which probably does the job of making a home personal much better than what a quarter inch drill can do. In words of Clay Christensen in his book : Innovator’s solution “The critical unit of analysis, while analysing a job, should be circumstance and not the customer”. Let’s look at Uber from a jobs to done lens. What is the job users are trying to accomplish by hiring Uber? Users are not hiring Uber to get them a cab. The job to done here is to transport users from point A to point B. Essentially cabs, e-scooters, trains, trams are trying to accomplish the same job for the user through different means — human driven car v/s an electric charged scooter. These different means have underlying different business model but the user doesn’t care about that. The user will pick whatever options that suits him/her best in that particular “circumstance”. Uber wants to be the product users hire to get them to any destination in all possible circumstances. Uber is Trying to Build Transportation as a Service Layer Irrespective of the circumstance you are in, Uber can give you best option to transport from point A to point B. In short term, this will lead to reduction in revenue for Uber. Essentially, all transportation options are competing against each other in this model on one platform. The long-term gain for Uber is a superior mobility experience for the end user that leads to sticky users and higher retention. The ambition for Uber should be to reach a point in future that the user experience of getting to any destination is so superior on Uber that majority of mobility transaction in a city starts happening on its platform. That way Uber can leverage its huge sticky user-base, the demand side, to start charging on a per transaction to the supply side — public transportation. The long term advantage here out-weights short term loss. Another aspect of doing this, from Uber’s standpoint, is building habits in users. Amazon has been trying to build habits in users for years. Whenever you think about buying something online you think of Amazon. Uber is trying to build same habits in users for transportation. It wants to be in a position that users don’t open Google maps to check how to get to a specific place but always open Uber for that. A caveat to be considered is that changing habits is even more difficult than building a new one. Uber has long way to go to instil a new transportation habit in users but it is on right track to do that.

Why Uber and Lyft Feature Competitors Who Could Cannibalize Them

More than a comparison between and , this is a set of thoughts and observations that might help understand the dynamics of taking over established industries. Uber Lyft innovative business models Indeed, when building up a company, a conventional is to design a business model and roll it up in the business world. Or to assembling it piece by piece by gathering feedback from the market and iterating on it. strategy In some occasions though, a business model to be successful requires the transformation of the overall society. When that happens, need to become the builder of ecosystems rather than just companies. For instance, going back to and , their are trying to change an essential piece of our daily lives: transportation. business models Uber Lyft business models Mass car ownership Both and started more like movements than companies. Indeed, they both had an alternative vision compared to traditional transportation systems. Uber Lyft For instance, is about “making it easier to live without owning a personal car. Achieving that goal ultimately means improving urban life by reducing congestion, pollution and the need for parking spaces.” Uber ultimate goal At the same time, Lyft started as a movement back in 2012, and its mission is to “improve people’s lives with the world’s best transportation.” While they’re going in the same directions, those companies share different core values. Where Uber values: Expanding access Delivering reliability Providing choice Aligning needs Being upfront Lyft values things like: Culture and values Brand authenticity A multimodal transportation They might be using a different wordings to deliver the same message. However, from those core values different cultures form, and these cultures are a critical element to the ability of those companies to scale up. There is a phrase attributed to Peter Drucker, “culture eats for breakfast.” Whether or not he said it, we can’t know for sure. But I think this phrase raises an important point. strategy Strategy, in general, might sit on the same table of culture. In some occasions, when you’re rolling out an entirely new , that demands societal adjustments, then culture does eat and for breakfast. business model strategy business modeling In and case, their beginning as a sort of movement didn’t spring up from a vacuum, but it was a bottom-up movement that came from the drawbacks caused over the years by mass car ownership. Some of those significant drawbacks are: Uber Lyft Underutilization of vehicles which are parked for most of the time The inefficiency of owning cars, which led cities to build massive parking lots in urban dwellings The cost of ownership is enormous compared to the benefits of a vehicle, especially within cities Those movements started from large cities, and it’s not by chance that places like San Francisco, Los Angeles, and New York were the places where those services got first rolled out. That brings us to the next point. The formation of new macrotrends Globalization, digitalization, growing urban populations and the growing precarity of jobs all led to the creation of new trends that helped and enter the transportation market, to grow quickly. Some of those are: Uber Lyft A growing number of people prefer to avoid ownership of cars The Internat has made it possible to use certain services on-demand The need for additional income has created a parallel underutilized “cheap job market” People’s growing environmental concerns are making them more aware of their transportation choices The inefficiency of public transportation and the taxi industry has in some cities spurred the need for new alternatives Those macrotrends undoubtedly helped and to roll out their , nonetheless regulation, and protests from existing industries (like the taxi industry). People pushed it for those companies success at a bottom-up level. Uber Lyft business models That also brings us to another critical point. When business models become ecosystems-maker While and , both have two key partners (drivers and riders) their business model success depends on the ability to deal with local communities. Uber Lyft Thus, the overall implementation of this new transportation requires the understanding of complex dynamics that help build ecosystems. business model Indeed, while if you’re trying to build a company in an established industry, all you have to do is to create a sustainable business model able to develop a thriving community. When you’re rolling out a on a whole new industry, that business model needs to plant the seeds to build a new ecosystem. business model Thus, rather than looking internally, it needs to look outside, and while it starts from building a community, it then needs to be able to scale an entire ecosystem! With new industries, new opportunities emerge It is interesting to notice how the emergence of players like Uber and Lyft are also allowing the growth of complementary industries. For instance, Hyrecar is a peer-to-peer marketplace where owners of a car can rent their idle vehicles to drivers that want instead make an additional income by driving for or . Uber Lyft By providing a that makes it easy for car owners and drivers to connect, Hyrecar is building a new business from scratch. platform Why is this happening? aren’t an easy game, and Uber and Lyft have come up with clever strategies over the years, mainly to keep drivers going back to the platform. Platforms Indeed, of the two key players (drivers and riders), probably the most important for them, to generate are drivers. If as a rider I get on the app, but there is no rider around, or my wait time is too long I might well switch to another service or transportation mode altogether. network effects For that matter, both and Lyft use a dynamic pricing : Uber strategy Lyft dynamic pricing to incentivize drivers to drive in certain areas Uber surge pricing to keep drivers going back to the platform to earn extra income Where Uber and Lyft are successful in attracting a broader number of drivers, companies like Hyrecar benefit from those network effects. Platform business models domination Uber and Lyft aren’t just companies; those are platforms. A at its core is a place that connects people and groups and makes their interactions as smooth as possible. platform As a result of allowing those interactions platforms to collect fees, which dynamics is not different from that of paying taxes in the long-run. That’s also why and Lyft are not in the business of owning cars. They allow connections that spur marketplaces. From those marketplaces, they extract economic value in forms of fees. Uber In the last two decades, the web has allowed many to be formed, Uber and Lyft are no exception. platforms Transportation as a service Another radical change that is going through is a broader transition from ownership to possession or use. In short, in many industries, including software (with the rise of ) companies are transitioning from offering products or one-time services to ongoing services that can be benefited via a subscription plan. SaaS business models For instance, Lyft introduced subscription fees to the access of the network of shared bikes and scooters on the or to get access to special promotions. platform As pointed out on Lyft blog with a subscription fee you can get an interesting transportation package: : Lyft Blog Source Many companies are trying to make the switch from product-centered to customer-centered, which implies a transition from single sales touch points to a continuous stream of services and recommendation for paying subscribers. Would this trend pick up, then it might become easier also for other businesses like Lyft to implement and improve their subscription offerings. Key takeaway New industries can be built on top of older and already existing ones. In the case of mass car ownership, more and more people have opted for a model where they don’t own a car anymore. This transition is part of a more profound movement, started almost a decade ago, and that is still ongoing. When new entrants come in and roll out their , it is critical those business models also take into account the fact that a whole ecosystem needs to be built if the business model itself needs to be successful. business models For that matter, companies like Uber and Lyft transition from platforms to ecosystem builders. Only when that process is completed, the will turn out to be successful, and the bottom line might be finally looking good! business model Originally published at fourweekmba.com on March 2, 2019.

Uber And Lyft’s Business Models in Laymen’s Terms

Like almost everyone else in tech, we were excited to see drop last Friday. The rideshare industry has fundamentally changed how millions of people (including us!) get around on a daily basis, which is an incredible accomplishment given that Lyft and Uber have been around for less than a decade. It’s rare to see a product reach near ubiquity over a relatively short period of time, and this feels like a unique opportunity to watch a transformational consumer tech company debut on the public market. Lyft’s S-1 We’ll save you a page-by-page analysis of the S-1 — instead, we wanted to highlight five of the most surprising things we read and what they suggest about Lyft’s business. Given that there aren’t public comps in the rideshare business, we’ll likely learn more about Lyft’s true financial health once Uber’s S-1 is public. But, we’ve compared Lyft’s progress to other public marketplaces when appropriate to give context for some of the company’s key metrics. P.S. — if you’re working on an early stage startup in the on-demand or marketplace space, please reach out at twins@crv.com — we’d love to talk to you! For the core ridesharing business, Lyft’s bookings are the total amount of money collected from riders. However, Lyft doesn’t keep all of the bookings as revenue — the company pays out a wage to the driver, so Lyft’s net revenue is (fare quoted to rider)-(wage paid to driver). For pure marketplaces, we often see 5–15% of bookings (or GMV) as net revenue — in of eight public marketplaces, we saw a median take of 14%. In 2018, Lyft saw a ~27% revenue take rate, up from 18% in 2016. 1. Lyft’s take rate (net revenue/bookings) of 25%+ is surprisingly high. our analysis Graphic created from data in Lyft’s S-1. Lyft’s take rate is increasing for two reasons — the company is: (1) giving riders fewer discounts, and taking a higher cut from drivers; and (2) paying less in driver incentives (bonuses to encourage them to drive). Lyft sees these trends as industry-wide — as ridesharing has matured, all platforms have lowered deals to get riders and drivers to engage. Lyft forecasts this take rate will increase, as they see more room for improvement in incentives. They also expect their bike/scooter business to grow (this revenue was non-material in 2018), which is 100% net revenue — Lyft owns and operates the vehicles. While Uber’s financials are not yet public, the company’s take rate last year was (or upwards of) 25% — suggesting high take rates may be systemic to rideshare. In a on marketplace KPIs, Accel’s Andrei Brasoveanu referenced the ability to charge a higher take if your marketplace is a “key distribution channel.” This is certainly the case for rideshare, as it would be almost prohibitively difficult for a driver to spin up their own network, particularly in contrast to a marketplace like Etsy where the seller could list items on their own website, at physical marketplaces, or in stores. reportedly close to great piece In conjunction with Upwork’s IPO last year, we analyzed other public marketplace take rates — which vary, but tend to be higher when a marketplace has a core distribution advantage. In some ways, it therefore seems fair Lyft and Uber charge a higher take — as well as acquiring customers for their drivers, they handle regulation, customer service, and insurance (we’ll get to this later!) — seller services that are a lot more intensive than those you’d find in a typical online marketplace. Lyft’s leadership has always been vocal about the importance of being a good corporate citizen, but we were surprised by how strongly this was emphasized as a competitive advantage. Frequent words/phrases in the S-1 included (10x), (5x), (4x), and (4x). 2. A “mission-driven brand” is portrayed as the way to beat Uber. social responsibility authenticity empathy mission-driven Lyft’s mission statement, pictured above in the S-1, is similar in concept but differs in tone from Uber’s mission statement (on the company’s website): “We ignite opportunity by setting the world in motion.” Lyft attributed its recent U.S. market share gains (39% in December 2018, up from 22% in December 2016) to this branding resonating with consumers. Lyft has certainly benefited from Uber’s stumbles in at least a few markets (remember ?), and the S-1 outlines Lyft’s conviction that “users are increasingly choosing a ridesharing platform based on brand affinity and value alignment.” #deleteuber Second Measure’s graphic above shows how Uber’s U.S. market share fluctuated during the company’s rocky year of 2017. Second Measure estimates market share using a sample of credit card data — while their 18% estimate for Lyft’s market share at the beginning of 2016 is somewhat close to the 22% stated in the S-1, they estimated that Lyft only had in January 2019 — compared to 39% stated in the S-1. 29% market share The S-1 also details some of Lyft’s specific actions and campaigns around corporate responsibility (present and future): 100% of tips go to drivers, with $750M distributed so far $10M donated to charities via Lyft’s Round Up and Donate program (from 700k riders) Free rides during natural disasters, emergency situations, and on voting days for underserved communities Rides are carbon neutral— Lyft is one of the largest voluntary purchasers of carbon offsets Partnership with cities to offer bike/scooter rides to low-income residents Plan to invest the greater of 1% of profits or $50M to social impact efforts annually Lyft’s gross margin is 42% (up from 19% in 2016), but the company’s EBITDA margin is -45% (up from -201.7%). These numbers are relatively low based on public marketplace comps, which typically see EBITDA margins of 5–10% — reflecting the fact that Lyft is much more complex than a pure software marketplace. 3. Huge insurance and incentive costs hit margins. By far the largest hits on profitability are cost of goods sold (58% of revenue in 2018) and sales and marketing (37% of revenue). Within COGS, insurance costs increased by $319M in 2018, compared to only $110M for payment processing, and $75M for platform hosting. The Lyft team details insurance needs at length in the S-1 — the company provides $1M in commercial automobile liability for each driver, which is managed through their own insurance subsidiary and third-party providers. Graphic created from data in Lyft’s S-1. As of EoY 2018, Lyft was holding nearly $865M in an insurance claims payment account (more than doubled from $360M in 2017). The S-1 attributes a significant increase in insurance costs in Q1 2018 due to “increased frequency and severity of claims,” though the company still appears to be somewhat cautious in estimating reserves. From 2016–2018, only 20–35% of the year’s reserves were paid out in losses. The second largest drag on margins is sales and marketing — which is where driver referral and rider incentive (discount) costs are allocated, which represented 36% of all sales and marketing spend in 2018. Lyft paid out $296.6 million in targeted rider and driver incentives in 2018, almost double the 2017 amount of $155.6 million (which was up only slightly from $141 million in 2016). Uber’s work in autonomous driving has been heavily publicized, especially during the company’s drawn-out lawsuit with Waymo over the acquisition of Otto. But Lyft is also betting big on autonomous vehicles, which were referenced more than 100 times in the S-1. Lyft, through a partnership with Aptiv (formerly nuTonomy) has completed nearly 35,000 autonomous rides (accompanied by a human safety driver) in Las Vegas. 4. Lyft has invested heavily in AV — and is closer behind Uber than they may seem. Lyft has completed 35k monitored autonomous rides through its partnership with Aptiv — image from Lyft’s . blog Of Lyft’s 4,700 employees, “several hundred” work in the company’s AV lab in Palo Alto, the Level 5 Engineering Center. Lyft plans to build its own AV system and partner with third-party developers who can build on Lyft’s Open Platform. Last year, Lyft disclosed its acquisition of computer vision startup Blue Vision Labs, as well as its with an auto manufacturer to build AV tech — both of which are referenced in the S-1. $72M five-year agreement As a result of these efforts, Lyft hopes to serve a “portion” of rides via AV in the next five years, and the majority of rides via AV in the next ten years. Uber, by contrast, has pursued a more aggressive strategy in deploying autonomous tech — after acquiring Otto for a reported $700 million in 2016, the company started AV tests in four North American markets (at least one without a DMV permit). Uber on deploying a driverless car service by EoY 2018, but has scaled back with new CEO Dara Khosrowshahi. Similar to Lyft’s projections in the S-1, Khosrowshahi stated in early 2018 that Uber’s ability to serve of a city’s rides autonomously is still 10–15 years out. originally planned the majority You may remember Lyft’s $250M acquisition of Motivate, the largest bikeshare company in the U.S., in late 2018. According to the S-1, the acquisition gave Lyft 200 new employees, as well as a commitment to invest $100 million into the New York bikesharing ecosystem over the next five years. Based on the frequency of mentions in the S-1 (“bikes and scooters” comes up 105 times), providing short-distance EV transport is crucial to Lyft’s mission of enabling multimodal “transportation as a service.” 5. Bike and scooter revenue is “not material” now — but likely will be soon. Image from Lyft’s S-1 — Lyft’s transportation as a service (TaaS) network spans local transit (info provided for free in the app), short-distance bikes and scooters, and a range of longer-distance vehicles. Lyft did not disclose bike and scooter revenue, beyond stating that it was “not material” for 2018 — which is unsurprising given the acquisition took place in November. However, Lyft’s disclosure in the S-1 that there is “no material difference” in active revenue per ride between rideshare and bikes/scooters gives us some more insight into this business line. Because Lyft has a 28% take rate on rideshare and a 100% take rate on bikes and scooters, this implies the average rideshare booking is ~3.5x the average scooter booking. Graphic created from data in Lyft’s S-1. Lyft does not outline revenue per ride in the S-1, but we calculated it at $2.80 given the rides taken and revenue earned figures for 2016–2018. Since Lyft’s bike/scooter share services have only been deployed since Q4 2018, this implies the average bike/scooter booking is ~$3. At a $1 cost to unlock and $0.15 per minute to ride, this implies an average ~13 minute ride. This is relatively in-line with data that has been by Bird, Lime, Jump, and other major e-vehicle operators, though we don’t know what Lyft’s vehicle utilization looks like compared to competitors. In the S-1’s, Lyft’s listed concerns about bike and scooter operations include: reliance on and potential quality issues from third-party manufacturers, uncertain weather-related seasonality, and “negative public perception” around e-vehicle launches in new markets. publicly disclosed Lyft did not disclose specific plans for expanding the bike and scooter service, beyond growing a pilot that connects riders to local transit or a Lyft shared ride. The company estimates the useful life of a scooter as 12 months for depreciation purposes (which seems somewhat optimistic, given a suggesting that the average lifespan is closer to one month), and as of EoY 2018 had $23M in value of scooters not yet deployed. In total, the company listed $68M of bike and scooter fleet acquisition costs for 2018. recent report Thanks for reading! We’d love to get your thoughts on anything you found to be particularly surprising or interesting in Lyft’s S-1 — feel free to email us at or tweet us . twins@crv.com @venturetwins Like this post? to help more people see our story, and share it with your friends! Clap for it Interested in reading more content from us? You can subscribe to our weekly newsletter, , for more insights on millennial and Gen Z trends: Accelerated https://accelerated.carrd.co/

Five Surprising Things in Lyft’s S-1

I was in New York City last month, in a Lyft, and I asked the driver how he was doing. He said it was a slow day. I’d heard that it was slow for the service industry in January. I asked him how much money he makes on an average day. Four hundred dollars. Four hundred dollars times 5 days a week times 52 weeks a year… is $104,000. He explained that he works for Lyft, Uber, and two other public apps. He also works for two other private app companies that allow him to pick up and drop off at the airport. He takes whichever job comes in first and off he goes. He’s working six different apps so he can maximize his time and his profits. He’s found a way for technology to work for him. Instead of waiting on one or two apps for a job to come through, especially in the slower months, he doesn’t wait. He goes from one app to the next so that he’s always on the go, always making money throughout his work day. He’s leveraging the technology to make the most amount of money in the least amount of time. That’s what the new internet is going to enable for everyone. The new internet will be layered with blockchain technology. The tokenization of all forms of value and the transfer of crypto will happen seamlessly, with minimal friction and miniscule fees. As blockchain technology allows for a more efficient means of value transfer, micro-payments will be made over the internet like credit card money transfers are now, only it will cost less and it will happen almost instantly, and in most cases there will be no big banks or interemediaries taking a cut. Soon, making $104,000 a year will become even more commonplace for anyone interested in leveraging their time through efficiencies inherent in the technology. The new internet will allow users to make money (by earning convertible tokens) — doing things we do now for free. Soon, and in some cases even now, we can make money by listening to podcasts, watching ads, writing blog posts, exercising, playing video games, and more things we do everyday anyway. If we’re going to do these things anyway, why not make money for doing them? Back to the Lyft driver I met. $104,000/year works out to $50/hour. That’s more than 3x the newly increased hourly rate for minimum wage and almost double the average American household income for 2018. There’s a new internet being built right now, and with it, new opportunities. It’s not just for the tech savvy. Making money over the new internet will one day be for everyone.

How One Lyft Driver Leverages Technology To Make Over $100,000 A Year

Tipping sucks, but we should still tip our ridesharing drivers. Here’s why. Last year, I drove for Lyft in the Denver area. I , discovering that under a third of my riders actually tipped me. To my surprise, this article received polar reactions on the issue of Uber/Lyft tipping etiquette. Many readers were floored that Lyft passengers don’t tip: recently posted an article analyzing my tips _Tips are as essential with ridesharing as they are for servers. I mean holy shit people, you’re hiring a private car…_medium.com As a fairly frequent rideshare rider, I’m SHOCKED that so many people don’t tip, ESPECIALLY for… Yet, most others defended their no-tipping practices: _I would rather they raise their rates and prohibit tipping._medium.com I’m sick of tipping people for doing their job. medium.com In my view, part of the advantage of using an app instead of a taxi is that I expect the tip to be incorporated into the fee. So, who’s right? What’s the appropriate tipping etiquette as a Lyft and Uber passenger? To answer these questions, I researched the origins of tipping, fair wage law, and the ridesharing industry. But let’s first understand how this whole tipping thing came to be. The Origin of Tipping By Soshial [ ], . Data collected from . CC BY-SA 4.0 from Wikimedia Commons Wikivoyage tipping sections Tipping today is a uniquely American phenomenon. Few restaurant bills outside North America command a pseudo-mandatory 15–20% service fee. In fact, Americans hand out $42 billion in gratuities every year — all without a law enforcing the payment. So, how did this now-ubiquitous practice begin? The first recorded instances of gratuity , where the aristocratic class could provide an “allowance” to the relatively poor service workers. A higher tip garnered better service, one reason that scholars surmise the word is an acronym for “To Insure Promptitude.” originated in Tudor England Flashforward to after the Civil War in the late 19th century. Wealthy Americans traveled to Europe, and brought back with them the culture of tipping. These Americans brandished their newfound habits to demonstrate wealth and worldliness. Tipping soon went mainstream. Many, like those of you who don’t want to tip Uber/Lyft drivers, recoiled at the idea of a tipping culture. The published an article where it described tipping as the “vilest of imported vices.” Titled a 1916 author wrote that those who depended on tips suffered from a “moral malady.” New York Times in 1897 “The Itching Palm,” Founded in 1904, the Georgia-based Anti-Tipping Society of America successfully lobbied six states to ban tipping. However, these laws went unenforced, and were all ultimately repealed by 1926. Today, tipping is an integrated part of the American economy. “Tip credit” legislation allows restaurants to pay waitstaff below the minimum wage, assuming an expected income from gratuity. Yet, most remain unaware of tipping’s elitist, condescending origins. Tipping Steve Buscemi’s Mr. Pink on why he doesn’t tip, in Quentin Tarantino’s “Reservoir Dogs” As is now evident, the age-old tipping debate is a complicated one. Most proponents of gratuity tout that tipping allows customers to incentivize and reward superior service. Others believe that tipping is an important element of control offered to purchasers, where they can course-correct for exceptional (or sub-par) treatment. Professor Michael Lynn Yet, the evidence doesn’t support these conclusions. has published over fifty papers on tipping. Lynn reports that the tip amount is rarely, if ever, correlated with the quality of service received. Instead, tip amounts increase for essentially random reasons, like if you pay with credit card or if your server writes a smiley face on the receipt. As an absurd example, Lynn says that even the weather conditions are more correlated with the tip amount than the quality of the service received. Professor Michael Lynn at the Cornell School of Hotel Administration Other opponents of tipping culture argue that tipping is a hotbed for unfair discrimination and sexual harassment. Last month, the from service workers “weighing harassing behavior against the tips they need to make a living wage.” New York Times highlighted personal accounts Even restauranteurs dislike tipping. Tipping creates an unfair ecosystem where service workers earn more money than cooks and kitchen workers, despite their often more skilled labor. Restaurant owners struggle to pay their non-service staff a comparably fair amount, and the imbalanced opportunity undermines workplace teamwork. When asked for his personal opinion, Michael Lynn, the guru on tipping himself, , “You know, I think I would outlaw it.” told Freakonomics Radio Okay, so Tipping Isn’t Great. But What Should We Do? Your initial response might be to stop tipping Uber and Lyft drivers. After all, tipping originated as an elitist, condescending custom. Tipping very clearly isn’t correlated with quality of service, and is the cause of several deleterious workplace practices. The advent of ridesharing applications has created a gray area for driver tipping custom. Uber’s initial messaging informed riders they are “not obligated” to tip, claiming Uber’s drivers made fair wages without the tip. Many early Uber adopters rejoiced at the idea of a no-tip taxi service, which became one of Uber’s selling points. Yet in 2017, Uber eventually added tipping features to its platform, reaching parity with Lyft. Screenshots from Uber promotional materials. Left is pre-2017. Right is after the tipping features were added. Uber’s initial rejection of gratuities has caused many to still refuse to tip their ridesharing driver. Some believe we can harness this ambiguity to shift American culture away from expected gratuities in the driving industry, and ultimately, to a fairer and more transparent economy. But, as much as tips seem to be something worth abolishing, this isn’t the right way. Here’s why. The Lyft and Uber Wage Debate In the past month, there’s been an intense debate about the fairness of Uber and Lyft wages. It started when researcher Stephen Zoepf published . The study was flawed in several ways, and received . a now debunked MIT study claiming Uber drivers earn $3.37 an hour criticism from Uber’s CEO After integrating the feedback, Zoepf , but they don’t look much better. In the , Zoepf claims make less than their state’s minimum wage at an overall median rate of $10/hour. updated his numbers best case 41% of Uber drivers Lyft, anticipating similar criticism, published showing that drivers make $18.83 hourly, without expenses. They claim expenses to be about $3–5/hour, which brings the hourly wage to $12.83-$15.83. This tipping income. an infographic includes Lyft’s infographic explaining wage hours. Note: The $18.83 number includes time that drivers spend seeking rides, not just the time they are actually driving customers. I interviewed Moira Muntz, the Independent Drivers’ Guild spokesperson, who found Lyft’s calculation to be misleading. Muntz points out that Lyft doesn’t deduct payroll tax, which is passed on to the drivers as self-employment tax, totaling to an even greater fee for the drivers. As , I believe these estimates still overestimate a ridesharing driver’s median income. To even sign up as a Lyft driver I had to pay over $200 for car maintenance work, car registration, and updated insurance plans. Now factor in the gas, cleanings, and extra niceties like water bottles and gum, needed to maintain a decent rating. I also didn’t receive common benefits like overtime, health insurance, and of course, minimum wage protections. a former Lyft driver myself So, are Lyft and Uber drivers compensated fairly? Are the hourly wages actually above minimum wage when considering these deductions? We don’t really know. Ok, it’s unclear if Uber and Lyft drivers are compensated fairly. Does this mean we should tip? Should we just tip anyone who makes below a minimum wage? Why should supporting fair wages be the onus of the tipping customer? Shouldn’t companies or the government make these changes? It’s a hard question. Muntz believes that “what drivers really need is for elected officials to regulate basic protections like a minimum pay rate.” But, in the meantime, there’s also an industry precedent here for tipping — Muntz says that “for-hire vehicle drivers have traditionally counted on tips to make up 20% of their income.” My answer: before abolishing tipping culture in the driving industry, we should push companies to guarantee a just wage system. Imagine a group of restaurant chains proclaim you no longer need to tip its waiters. But the waiters’ wages don’t change — they’re still paid under the minimum wage. Few would observe the restaurants’ new rule, since many would feel compelled to tip out of a moral duty. Uber’s tipping ban was built on what now appears to be an unverified claim — that its drivers make fair wages. The pre-existing gratuity culture for drivers shouldn’t be changed until we can actually ensure fair compensation. If you believe in fighting tipping culture, there are better ways to do it than by continuing to underpay a low-income job for many who are trying to make ends meet. If you blame the industry itself for being corrupt in its business practices, then why take Uber and Lyft in the first place? Instead, you can take steps in the right direction to change tipping culture and uphold fair ridesharing driver wages. Here’s a few ways: Encourage legislation that regulates fair minimum wage for ridesharing drivers. Muntz suggests folks sign the . IDG’s pay petition Spread awareness of the unfair wages that Lyft and Uber drivers receive. Support organizations like and that are fighting for ridesharing driver rights. Independent Drivers Guild United Private Hire Drivers I’m all for eradication of tipping culture. But let’s intentionally and alter these economics. fairly Tipping isn’t an easy issue, and causes many people undue social stress. This article is my opinion, based on my experience as a Lyft driver, but I hope you learned some interesting history about tipping as well. You can check out some of my other most popular articles below: I hacked HQ Trivia But Here’s How They Can Stop Me How I Read Books So They Last How I Live With 115 Items

So, Are You Supposed to Tip your Uber/Lyft Driver?

Lessons from the road as a former Lyft driver in the Denver area. For rideshare drivers, tips are an important source of income. But what variables correlate to a large or small tip? Do controllable factors like conversation or the day of the week affect tip outcomes? I designed a mini-experiment to investigate. After each ride, I collected selected information on the ride, including the tip amount I received. Of the fifty passengers in my data set, sixteen of them tipped. Disclaimer: This experiment is mostly for fun and a personal experience. Rigorous analysis requires further data collection. Data Collection These are the variables I collected for each ride: Time start and time end (to compute ride duration) Day of week Number of passengers Gender of passenger with the rideshare app Conversation? (yes/no) Type of Lyft (Lyft vs. Lyft Line vs. Lyft Premier) After each ride, I recorded these variables, without any other related passenger information. The Analysis I performed some back-of-the-envelope analyses to observe significant correlations in the data. I found only two of the above variables to be significantly correlated with a tip: conversation and the type of Lyft. Conversation Almost all my rides with tips came from conversing passengers. However, I had conversations with many passengers who did not tip. The conclusion: passengers who tip nearly always converse, but it isn’t necessarily the case that all passengers who converse are those who tip. This result makes sense; a passenger may not feel a desire to tip if they had an ordinary ride experience and didn’t interact with the driver at all. Type of Lyft As might be expected, the type of Lyft taken was an excellent predictor of whether or not the passenger would tip. Zero of my ten Lyft Line passengers tipped me. This effect is amplified when you consider that most Lyft Line rides have multiple passenger parties — in theory, there should be more opportunities to earn a tip. My interpretation: the cost-conscious Lyft Line passengers are less likely to tip. Although Lyft promotes Lyft Lines to drivers as an opportunity to earn more tips from the additional passengers, my personal experience doesn’t match up. Other Insights: Tip Distribution Above is a chart showing the distribution of my earned tips as a driver. Other than the most common scenario of not getting tipped, tips were most often either $1 or $5. This likely has to do with Lyft’s passenger UI, which provides tip pricing suggestions for $1, $2, and $5. Tipping UI screenshot from Lyft.com The Conclusion More data collection is needed to draw insights. A huge bias in this dataset is the fact that all fifty rides were driven by a single person (me), so it’s hard to extrapolate this data out to a larger population. After investigating the data, I have two hypotheses for further study: Most rides where Lyft drivers get tipped are ones where the driver and the passenger had a conversation. Most tipping passengers are taking Lyft or Lyft Premiers, not Lyft Lines. It’s worth noting how different the tipping culture is for Lyft rides than for taxi rides. While the vast majority of individuals tip 15% for a taxi ride, under a third of my passengers even chose to tip me. With all , I hope people consider this difference in earned tips and its effect on driver income. the recent discussion about whether or not Uber and Lyft undercharge drivers PSA: Want to support fair wages for Lyft and Uber drivers? Join the Independent Drivers Guild (IDG) Petition . The IDG is a Machinists Union affiliate representing and advocating for more than 60k app based drivers in NYC. here

I analyzed my Lyft driver tips, here’s what I found

For a month last year, I drove for Lyft in the Denver area. These are my stories. It’s 10:03PM. I pull up to the one-story house, and wait four minutes in the darkness. The passenger is nowhere to be seen. But then, I see a young, mid-20s woman emerge from the front door. Whispering, she asks, “Can I fit some boxes in your car?”“Sure, are you going to the airport?”“No, my boyfriend broke up with me and I have all my stuff here and, and…” Tears stream down her face and her voice cracks. “No problem!” I quickly respond. “Take as long as you need! Let me know if you need help!” The Lyft app notifies me the waiting time has expired and I must move on, but I ignore it. After another ten minutes, I see boxes piling up on the front porch. We move her life’s possessions to my now completely full trunk. Meanwhile, a few people step out of the house to watch. A tall lanky man in a wifebeater helps her move the luggage, apologetically. Two children watch from the cracked four-paned window. As we hit the road, she calls someone for a few minutes and half-speaks half-wails in a foreign language. I offer a couple platitudes, out of a loss of what to say, and she nods. Silence for the rest of the ride. We arrive, and I help move her boxes inside the house, to what looks like her family. As I depart she says, “Thank you.”

Lyft Confessionals: a woman kicked out of her home

On May 1, 2017, Lyft announced its program. Designed to strengthen the community Lyft was built on, and support a variety of nonprofits, the allows riders to “opt-in” and have their fares rounded up to the nearest dollar so that the change can be donated to a charitable cause. Round Up & Donate program Since its inception, the Round Up & program has been a massive success. In only seven months, the program has raised more than $3 million. While it seems obvious that riders would want to participate in such a wholesome program, has surprised many. Donate the success of Round Up & Donate While it may seem like magic, Lyft’s Round Up program creatively leverages a few behavioral economic theories to make donating pain free and seamless. Why Lyft’s New Program Works: 3 Reasons Rooted in Behavioral Economics Lyft’s Round Up program wasn’t an accident, and its massive success isn’t random. Instead, Lyft’s engineers utilized three essential components of behavioral economics and built the program around it — whether they knew it or not. Taken together, these three ideas make the concept of rounding up attractive, easy and exciting for customers, reducing the friction between intention and action. 1. The Default Effect In the world of behavioral economics, the is the course of action that happens when the decision maker doesn’t make any specifications. default effect Because it’s easier to just go with the default setting, people tend not to cancel subscriptions or opt-out if they don’t want to donate. Lyft leveraged the default effect with its Round Up program. Instead of asking the user to donate to the nearest dollar after , Lyft asks you to set it as the default and then auto-donates after every ride. The program runs in the background of the Lyft app, and riders almost forget it exists. every ride 2. Anchoring is another tenant of behavioral economics. It refers to the cognitive bias that happens in relation to the first piece of information a person gets when making a financial decision. Anchoring In other words, it’s the human tendency to evaluate an option based on the other options present. Here’s how it works in Lyft’s case: Say a rider is already spending $8.20 for a ride. At that point, what’s another $0.80? Most people opt-in without even giving it much thought. Much like an upsell, the Round Up program works because the hardest part is always getting people to spend . Once they’ve spent some money, it’s pretty easy to get them to add a few extra cents. any money 3. Framing refers to the way an organization words a choice to make it appear more or less attractive to customers. On its website, Lyft states that customers who opt-in donate “just a few dollars each month,” on average. Framing As you’ll notice, Lyft doesn’t ask riders to commit to donate money outright every month, which wouldn’t be nearly as successful. Instead, by using the phrasing “round up to donate,” the company makes the ask seem simple and straightforward, thereby making it much more attractive. Lyft Continues to Advance the Program The Round Up & Donate program continues to thrive, and Lyft recently added three new partners to the system. According to , the partners include Girls Who Code, Habitat for Humanity and the World Wildlife Fund (WWF). In the past, Lyft has also offered to match donations made via Apple Pay, resulting in more donation dollars and outstanding program performance. TechCrunch Until next time, Nate

The Secret to How Lyft’s Round Up Program Raised Over $3 Million

Uber Driver app running on Mac while I write this post. Driving for or can be a great source of income, whether it’s your primary job or side hustle. It is not uncommon for drivers to head over to San Francisco for a weekend and come home with a few thousand dollars. Even though I drove a work van in the city for years and can navigate it well, this “weekend warrior” strategy is not for me. Instead, I utilize one little trick that has paid off tremendously and drastically reduced the amount of hours I drive. Uber Lyft This strategy may not apply to everyone given the following circumstances, but it works extremely well if you live in a major (or even semi-major) metropolis: I live in a popular downtown area The area is relatively small It includes a 20,000 capacity arena with all sorts of events With a grid as small as Sacramento’s downtown area, it’s hard to make good money without having to make 10,000 small $3 trips. The average fare from one side of the grid to other is about $6.50, leaving the driver with nothing more than a few bucks. Learning this early on forced me to think of better ways to utilize my time. The trick is simple: plug your iPhone in (or ), mirror your Uber Driver app to your computer and wait. stream your Android To mirror your iPhone on a Mac: Plug your phone in with a lightning cable. Open Quicktime Player. Go to File > New Movie Recording. Change the input source to your iPhone. Change your Quicktime movie recording source to iPhone and open the Uber Driver app. You’ve now got a perfect view of your area inside the Uber Driver app displayed on your Mac. Feel free to browse the web, do some writing or catch up on work. Since there’s no real lag between the iPhone and Mac, you’ll see surges in near real-time. When it hits, turn your app on, wait for a ping, head out and make money! This strategy has allowed me to cut back on the amount of hours I drive while retaining high payouts. Ignore the poor design — I kinda suck at . Google Data Studio If you’re interested in driving for or , consider signing up using my links. Uber Lyft We both get a bonus if you do! I write other stuff like this at . I also curate a brief, that features articles similar to this from around the web. Nerdy Minutes weekly newsletter

How I Reduced my Uber & Lyft Drive Time and Earned More

An interesting message popped up on my radar from Kevin Law on September 21st, 2016: “You’d be a perfect fit for a company I’m advising if you’re extremely flexible on location” Since I love working on early consumer products and traveling, it was only three weeks later that I touched down at the airport in Kampala. Serve as interim CTO for 1 month and provide leadership to a small ride share engineering team, paving the way for future success. The mission: (Note — there are several HD YouTube videos of Uganda at the end of this post if you’re like me and you like to watch videos.) Boda Boda A “boda boda” is an inexpensive motorcycle ride offered in African cities (for my San Franciscans, imagine a ride from Hayes Valley to the Marina for just $1 versus a $5 Uber carpool). It is estimated that 1,000,000 boda rides take place everyday in Kampala. While on a walk, upon reaching a sidewalk, you will find you are solicited by a boda driver within seconds. “Boss!” “Boda!?” A regular boda ride involves hopping on the back of a boda driver’s low power motorcycle. He will take you on a harrowing ride, dangerously weaving through the chaotic gridlocked streets, sometimes crossing a median, often narrowly avoiding a collision or sideswiping a bigger vehicle, and sometimes just driving on the sidewalk. Passengers rarely have helmets, they sometimes double or triple-up on the back of a single bike, drivers are often unlicensed, and, at least in Kampala, there are very few traffic signals, street lights, or stop signs. At the end of the boda trip, you negotiate with your driver and you pay in cash. A look at the streets of Kampala during the day. Video shot . here Don’t let the above-mentioned video fool you- if it looks fairly organized in the video, one must only travel a block or so before you hit points which seem totally chaotic. Most intersections are very disorganized and seem extremely dangerous to a like me. mzungu Drivers Want “SafeBoda” The name of the company I consulted for is called . Its business hypothesis is that consumers may want a safer, organized, and vetted system for requesting a boda ride. SafeBoda drivers are screened, have a bright orange reflective vest, and have an extra helmet available for you to wear on the ride. The SafeBoda driver’s unfair advantage? - they get many more boda passengers than before they were a “SafeBoda” driver and therefore they make more money. SafeBoda drivers also have access to insurance, quality equipment, training in first-aid and road safety, and a sense of community that is rare in the industry. SafeBoda Evidence of early traction is that with virtually no technology, SafeBoda was able to sign up over 1,300 Kampalan boda drivers who each pay a fee as part of their onboarding. SafeBoda was so successful at this that they stopped accepting new drivers in order to focus on passenger demand and product. Before my arrival in Kampala, SafeBoda had a basic app in the Google Play Store with which around 1,000 rides per month were taking place. Clip of my “SafeBoda” ride from my apartment to the office Uganda’s Mobile Phone & Payment Technology Android is king in Uganda with around 70% market share. iPhones are rare… perhaps a 3% share. There are also ‘feature phones’ which only allow calls, texts, and a basic interactive textual menu provided by the carriers ( ) which I would estimate is used by the remaining 27% of users. USSD The major cellular carrier is called . In addition to data plans and voice calls, MTN also provides a money transfer service called “Mobile Money” which is tied to your SIM card and used by around 85% of Uganda’s citizens. MTN A Ugandan can deposit his or her pay (which is usually disbursed as cash) at any one of thousands of MTN locations throughout the country. MTN Money locations are usually offered as an extension to a physical store front such as a market or street vendor. After depositing cash at an MTN Money vendor, the balance on the user’s mobile interface instantly updates. With the updated balance on the mobile interface, the MTN Money balance can then be used to pay bills and transfer money to others at a cost of ~3% in fees on each transfer. Clearly as part of a technology revamp, SafeBoda should include MTN Money as part of the payment solution. However, we wouldn’t be able to complete the integration within 10 days due to MTN’s stone aged SOAP-based API. Instead, we decide to use Stripe’s API platform which we can implement in 1 day and which the founders felt would help to validate whether or not credit card holders are willing to use the service. It is decided MTN integration will take place in a subsequent development cycle. SafeBoda’s Legacy Tech Upon my arrival, the 5 person engineering team had first created a PHP / MySQL backend on 1 server, but then decided to rewrite it using with after they had encountered a series of problems related to connection issues and deliverability problems with the Android client. CodeIgniter Digital Ocean Node.js Socket.io On my arrival, some of the problems the local engineering team was grappling with were: When the client would miss a socket.io message, the app would strand The MySQL database was hard to migrate The code was peppered with large, hand rolled complex inline SQL queries The developers were actually testing against the production servers There was a lot of opportunity to improve both the development process and the product. Overhaul In 10 Days I convinced the founders to let me drive a complete overhaul to their product architecture and process in 10 days flat — roughly 300 man hours split between just 3 engineers. Having worked for over 2 years engineering at Lyft, I also know a thing or two about the best way to design almost anything related to ride share services. Scaleable product architecture and fast development process using technologies the team is already familiar with was right up my alley and my experience would allow us to leverage everything I had learned. I wanted to use Python and , but after interviewing the team, I was able to see that due to their lack of significant Python experience, there would be little chance we would have a working revamp in 10 days. I had to keep them on PHP… something that made sense for them since scaling wasn’t an issue yet and rapid prototyping and market validation were the main focus. We would generally build with an eye towards future scalability in case things explode out of the gate, but for now, fast development was priority #1. Flask With 2 of their existing devs, the three of us started a 10 day process in which we locked ourselves away in a room, worked almost until midnight every night, and emerged at the end of the 10 days with a new minimally viable system. Below is an animated gif (pardon the low resolution) of the “Golden Path” of a ride request & boda driver pickup. This recording was made on the evening of our 10th day of development. This screen recording used the real API’s that were working on the 10th night. A “rough cut” of the working app. All of the screens used real API’s behind the scenes on our new server architecture. Almost all of the values were dynamic and correctly pulling from the REST API. Obviously several ui features were not completed at this stage. The following week, the Android interfaces were updated: HD video of the golden path about 1 week after the animated gif was recorded. How We Made It In 10 Days When you rapidly build something, you start with an idea of what the features will be, what platforms it should run on, and who will do what. Given Android is the market leader, our mobile developer would build out all of the front end screens according to our product spec. We mapped out all of the screens on a whiteboard and then our Android dev was off to work. The other developer and I would would hack out the backend architecture. We would create a new development process, infrastructure, product architecture, and API’s. We would document them, and hand them to the Android dev to implement on the 4th day. Hosting on Heroku Hosting this project on propelled us faster and further than I had originally imaged. Using Heroku pipelines with GitHub integration meant that once a pull request’s tests pass in Circle CI, it automatically deploys into our staging environment. I had never used Heroku pipelines before, but I think it’s amazingly easy to setup and very powerful. Heroku Heroku’s awesome pipeline interface used for deploying to production The best part is that as we open up pull requests, Heroku senses it and offers us an ability to launch a test app called a “review app” — you see it above in the left column. One click will launch a fully operational environment that you can connect the mobile app to in order to test an upcoming feature. Setting all of this up takes only 3–4 hours, which includes learning how it all works. Dockerized Development In production, we use mlab.com to host MongoDB version 3.2. Docker allowed us to launch a local Docker development database container with 1 command: docker run — name mongodb -d mongo:3.2.10 (Note — some of the additional switches we use are omitted for brevity) As for PHP, I was able to create a PHP dev environment container that closely mirrors Heroku’s environment. The devs can do a similar docker ‘run’ statement to launch a local development container through which they can work on their local PHP files, do live debugging with PHPStorm via both web and cli requests, and even test their local development using an Android phone across an ngrok tunnel. PHP Application Framework The closest thing PHP seems to have to Python’s Flask is called . Lumen gives you a basic routing system and is setup to support REST API development. It makes versioning your REST API very easy. It also supports adding “Middleware” which performs things like API authentication. One thing I personally like is the IOC injector that detects which objects your controllers would like to use and automatically injects them for you by default. Lumen For Mongo access, we used PHP’s — an open source data mapper for MongoDB that we used back at and a project to which I have also been an . Doctrine ODM Lyft open source contributor to Gameplan On day 1, we mapped out our data model. We really only needed two key domain models: User & Ride. We then went to the whiteboard and defined our ride state transitions. The happy path looks something like this: ‘requested’ -> ‘accepted’ -> ‘arrived’ -> ‘pickedup’ -> ‘droppedoff’. There are two edge cases. One is that if our dispatcher fails to find you a ride, we set the state to ‘operatordispatch’ - a final state in which the mobile client simply displays the number to SafeBoda’s human powered dispatching call center. They’re able to manually setup dispatches to drivers who have feature phones. The other case is when the passenger cancels — that’s a final state where the request ends. If the driver cancels, the dispatch loop just continues. Below are some of the actual whiteboards drawn in our kickoff on day #1. Initial diagram of the ride initial state changes. Some chicken scratch from that first meeting where I described the way the phones would poll the server and how we would structure the HTTP JSON server response. From left to right: Charles, Kirk, and Jon (me). RestAPI We next sketched out a basic HTTP REST API that has methods such as: Create a ride request. POST /v1/ride Update a user’s current location. PUT /v1/userlocation In total, we have about 20 REST calls that were defined. We wrote the whole API spec up in an swagger.yml file that is rendered beautifully if viewed with . http://editor.swagger.io/ For the following 9 days, we hacked away. We didn’t attend any of the company meetings and we spent some of the time in my personal apartment so as to not be distracted by anything. Service Oriented Architecture As with most minimally viable apps, we created a tiny so-called ‘monolith’ app. I like to organize business logic into service classes — these become the foundation of what you would eventually ‘decompose’ later into application partitions, or what some people call “microservices”. In our app, we have the following PHP “service” classes which can easily be broken out into the future and implemented in any programming language into microservices: Behaviors responsible for converting between USD and UGX using the latest exchange rate. Currency Behaviors responsible for finding the nearest drivers, and powering the dispatching lifecycle. Dispatch Behaviors used to determine the number of minutes it will take to travel on the road between two places Eta Behaviors used to authenticate a Facebook user and retrieve profile data. Facebook Behaviors responsible for tracking analytics event information. Metrics Behaviors related to billing users using a payments gateway. Payments Behaviors related to pricing rides. Pricing Behaviors related to casting a rating for a ride and then also updating a driver’s record to update their average rating. Rating Behaviors related to creating and updating a ride. Ride Behaviors related to user creation and update. User Behaviors related to updating the user’s location. UserLocation Coping with Eventually Consistent Data Due to the fact that MongoDB is not an ACID database, the application must be able to cope with database references which can be out of sync. In MongoDB, there are no transactions or rollbacks. To help overcome concurrent update contention and data loss, we use optimistic record locking by default on all updates. We use Doctrine’s event hooks in order to ensure two records point at each other. For example, If I put user A in ride B, I will first update user A’s record to have ride_id = B’s id. Then, during the post update hook, I will next update ride B’s user_id to have user A’s id. There are short moments when the database is therefore out of sync during normal operations. Further, the app might encounter a show stopping error before the 2nd update hook completes, leaving the database in a state where the user record indicates the user is in a ride, but the ride does not correctly point back at the user and this would be the source of trouble if your code is not setup defensively. To cope with this, we must have protective ‘cleanup’ routines that check for situations like this. Before key business logic fires, we invoke a “cleanup” routine to ensure the data integrity is correct. Note that we do not employ MongoDB’s “ref” feature — we manually store record id’s and resolve database references manually. Also noteworthy is that when we update the user’s location on their user record, we perform the update without the default optimistic locking record strategy used throughout the rest of the software, or else we would see tons of optimistic lock exceptions for no good reason. 3rd Parties Tools We Used GeofenceAPI PagerDuty PaperTrail for MongoDB hosting Mlab.com for Jenkins & S3 hosting Ec2 for the API cert, for the Jenkins cert RapidSSL LetsEncrypt CircleCI Mixpanel InstaBug Features & Hindsight A surprising number of drivers in Uganda cannot read maps. Some clearly had trouble knowing left from right. The society is also highly cash based and most folks don’t have credit cards. If I had known some of these things, I would have advocated for a slightly different MVP in which relied less on screens with maps on them and put more emphasis on calling the driver. Conclusion The developers are no longer testing against production machines. We ditched complex hand rolled inline SQL for a MongoDB solution which requires no production migrations. If the project takes off, MongoDB will allow us to add more shards and rebalance data. It’s now up for the market to decide if the product is the right one for Uganda. A Personal Note Living in Kampala for a month and working on a real startup again was awesome. I feel personally enlightened and the trip gave me a new perspective on the world. In San Francisco, we have a lot of millionaires & billionaires and I often feel like I’ve missed out on something that everybody else was able to achieve. However, my trip to Africa made me realize just how fortunate I am to be a self made engineer and what an achievement it is to even be able to live in San Francisco in the first place. My 30-day contract has ended. I wish the founders the best of luck. Here are the current state of things: SafeBoda is seeking an on-site full-time CTO. If you’re up to the challenge, contact team@safeboda.com The version of SafeBoda in the Android Play Store is still the legacy version of the app. They continue to make iterative improvements to the upcoming platform, and have a strategy to release the new platform in January. The work we accomplished currently lives in the beta channel. Make me proud guys! Twitter: Email: block.jon@gmail.comJoin me on the DoorDash API Platform Team and browse other available positions: Contact: @blockjon https://www.doordash.com/careers/ Pictures And Videos From My Trip KLM flies from SFO all the way to Kampala’s international airport with layovers in Europe My former office: San Francisco’s luxury “China Basin” building. My Kampala office? This inexpensive building. Photo taken . here HD clip of a ride down a street in Kampala shot from an Uber. A look at some Rhinos which were going to sleep for the day World’s most powerful waterfall Heading towards the center of Murchison National Park — home to thousands of wild animals Giraffees in the road in Murchison National Park Elephant hanging out at the water’s edge on a hot day. Hippos in the water. Walking from my apartment to the coffee shop. Filmed . here A Ugandan lunch: Rice + G-nut sauce, sweet potatoes, cabbages, and greens. Wiping out in a Jinja class 4 rapid (me in front left) A sophisticated fashion event Buying a breakfast sandwich “Rolex” on the side of the road in a slum. Video shot . here Eating tilapia at the shore of Lake Victoria. Video shot . here Inside the ultra-modern Acacia Mall food court Buying a random rolex on the side of the road heading from Jinja to Kampala

How engineering @ Lyft helped me launch a new African ride share app in 10 days

, which in January organized an investment round in which General Motors participated with $500 million, , a scenario that contributes to delineating the following frontier in the industry: people are buying fewer automobiles, and that transport will no longer be about ownership but about services provided by third parties. Lyft has rejected a full buyout offer from the automotive company automobile The main players are now taking their positions: Daimler has created with Europcar, while the taxi app market has been consolidated through the merger of Hailo and MyTaxi, while BMW has launched with Sixt, Tesla says its autonomous vehicles will be able to “ ” transporting other people around when their owner does not need them, GM is interested in at the same time … it’s clear that this game of musical chairs is speeding up. Car2go DriveNow work for their owner Lyft as it launches Maven It’s becoming clear that the consequences of which we have been paying for too many years now. The idea that each family owns a number of vehicles that are used for about 3% of the time and usually only with one or two people aboard, using a ton of metal that is often no more than a status symbol to move one person, is absurd, and above all unsustainable. the age of the car was a historical mistake We have to start thinking that owning a car makes no sense and simply costs too much. Redesigning cities for people, getting rid of parking lots and spaces that could be used for other things, and instead opt for efficient transport. , and is … The next experiment in eliminating private cars from the city will probably be no less than Manhattan far from being the only one attempting to do so General Motors’ move makes sense: Lyft is no threat to Uber, but has valuable experience in the management of these types of services, despite its very limited development in a few US cities, but could still play an interesting role with some investment to boost its capacity. With Uber in the lead, Didi Chuxing out to conquer the world and other companies looking for partners, the idea of taking over a company like Lyft to explore the car-sharing market shows that these types of strategies are anything but testimonial: the automobile companies could opt to take over these companies to try to slow them down, but these seems unlikely when other companies are aggressively taking positions. After rejecting GM’s offer, Lyft has announced a new financing round, but the trend is clear to see. It is even possible that your car will be the last you own, and that after that you will be a satisfied user of a pool of vehicles that are there when you need them and then disappear when you get where you’re going. And not too far down the road, you won’t even be driving them, they’ll be self-driving. If the carmakers have cottoned on to this, then there really isn’t much left to say… (En español, aquí )

GM and Lyft: the changing face of the automobile industry

Looks like it’s part 2 of my series on Data Quality! In a , I explored Airbnb’s strategy for enhancing data quality through incentives. They implemented a single score and clear scoring criteria to establish a common understanding between data producers and consumers, fostering a genuine sense of ownership. previous post https://hackernoon.com/data-quality-score-one-score-to-rule-them-all?embedable=true Now, Lyft is taking a distinct approach, not attempting the same thing differently, but rather focusing on different aspects of data quality. And, Lyft’s strategy complements Airbnb’s efforts. While I regard Airbnb’s DQ score (or any similar score) as an effective means to consolidate various attempts at elevating data quality, Lyft is tackling this challenge from a different angle. The Airbnb’s DQ score serves as a valuable tool for providing a concrete visualization of data quality. In essence, any initiative to enhance data quality should manifest a discernible impact on this score. On the other hand, Lyft presents one possible initiative to proactively enhance quality by testing and validating data against specific quality criteria. https://eng.lyft.com/from-big-data-to-better-data-ensuring-data-quality-with-verity-a996b49343f6?source=post_page-----415baad6c95b--------------------------------&embedable=true Fundamentally, it’s a different point in the data quality lifecycle. Introducing a mechanism to improve quality necessitates the ability to measure it initially. So, while ’s focus lies on and data quality, relying on the producer’s interest to enhance this quality and “look good,” Lyft takes a different approach. Lyft places emphasis on actively and data quality, providing both producers and consumers with the means to effectively improve and control the quality. Airbnb measuring observing testing validating Collectively, these approaches provide a comprehensive strategy to address and enhance throughout its . data quality lifecycle For this reason, I was particularly interested in taking a closer look at Lyft’s approach. Another factor that intrigued me is testing, more specifically, contract testing, which has been used for many years now in basic software engineering with the emergence of microservice architecture. Data contracts, however, are something more recent in the domain of data engineering and are seen as the pinnacle or one of the ultimate steps to put in place on the path to building high-quality data pipelines. This is why I wanted to examine Lyft’s approach in more detail and explore some potential parallels. As mentioned, the approaches taken by Airbnb and Lyft are complementary and aim to achieve the same goal: improving data quality. has developed the DQ score, which focuses on measuring and enhancing 4 distinct : Airbnb aspects of data quality DQ Score has guiding principles, including full coverage, automation, actionability, multi-dimensionality, and evolvability. It has dimensions like Accuracy, Reliability, Stewardship, and Usability. Lyft’s Verity is a platform designed to enhance data quality across 5 dimensions Defines data quality as the measure of how well data can be used as intended, covering aspects like . semantic correctness, consistency, completeness, uniqueness, well-formedness, and timeliness It’s easy to draw parallels between the 5 data quality aspects improved by Lyft’s Verity and the 4 data quality dimensions measured by the Airbnb’s DQ score. For example, aspects like Timeliness would certainly contribute to the DQ score’s Reliability, while Accuracy would be dependent on semantic correctness, completeness, and uniqueness of the data. The Usability score, on the other hand, is influenced by the consistency of the data among other factors. Lyft’s Verity focuses on defining related to . It follows a test-first and validation approach, whereas Airbnb’s unified DQ score emphasizes evaluating data quality through various dimensions. checks semantic correctness, consistency, completeness, uniqueness, well-formedness, and timeliness If we wanted to incorporate the DQ score into this last schema, it would be on the side of the Alert/Debug steps. Airbnb’s DQ score uses different signals to assess data quality across the . Accuracy, Reliability, Stewardship, and Usability aspects We also had a set of input signals that more directly measure quality (Midas certification, data validation, bugs, SLAs, automated DQ checks, etc.), whereas others were more like proxies for quality (e.g., valid ownership, good governance hygiene, the use of paved path tooling). As discussed earlier, there are likely overlaps between Airbnb’s DQ score and Verity. While Airbnb focuses on pushing data quality to the right, emphasizing measurement and scoring, Lyft’s Verity takes a proactive approach by shifting the check definition configurations, testing, and validation processes to the left, emphasizing proactive improvement of data quality. Now, my primary interest lies to the left, in the check definition configurations, testing, and validation aspects. How Lyft integrates data quality testing into its processes? Let’s first examine the execution paths. Currently, Lyft’s Verity is primarily focused on ensuring the quality of data stored in its Hive data warehouse. However, there are plans to expand its capabilities to support other data sources in the future. Note that while they refer to Hive as a data workhouse, they actually utilize it as a hybrid data storage solution, operating as a data warehouse for structured, processed, and cleansed data (silver layer) while also serving as a data lake for raw event data (bronze layer). Poor data quality in Hive caused tainted experimentation metrics, inaccurate machine learning features, and flawed executive dashboards. Verity’s checks can be integrated into an Airflow DAG to ensure that only high-quality raw data is processed and stored in Hive as derived data. Data producers and consumers can define their data quality checks and verify the data when it is produced or before it is consumed inside or . Airflow Flyte … The VerityAirflowOperator can be used in a blocking fashion to halt a DAG upon a check failure, preventing bad data from ever reaching production. This utilizes the “ ” pattern: we create data in a staged schema, verify the data with a blocking operator, then promote it to production if it passes quality checks. Stage-Check-Exchange Checks can also be performed manually or scheduled automatically to verify both raw and derived data. Verity Scheduled Checks are isolated from any data orchestration engine, so they still run even if Airflow or Flyte are completely down. This remedies a common problem of checks not alerting because the Airflow Task never ran. So, there are essentially 3 primary ways to trigger a check: as part of an Airflow DAG, manually, or scheduled through the Verity platform/UI. I don’t believe that current checks can be integrated into real-time streaming pipelines (such as Flink + Kafka) to validate for example data as it enters Hive (or even earlier). Implementing this type of real-time check would enable prompt detection of discrepancies, leading to reduced storage and processing costs and overall data quality enhancement. Well, to be thorough, Verity checks are managed through an API server, which could be used to trigger checks programmatically through some APIs. Verity API Server — This service handles all the external APIs regarding running checks as well as persisting and retrieving their results. The API Server does not execute any checks, but rather writes a message to our Check Queue, which utilizes (SQS). SimpleQueueService So, maybe, you could potentially trigger these jobs in a more real-time fashion, such as from a streaming job or even, as a long stretch, by integrating with Hive change data capture (CDC). However, when executed outside of Airflow, these jobs would not be able to block the data processing job; instead, they would generate asynchronous alerts pushed to the Check Queue. Some consumers would prefer to have data processing delayed when a check fails, while others would rather proceed and receive an alert. Now, let’s look at these data quality tests. Here’s an example that checks whether is never null. This is accomplished through a combination of query and condition components. rider_events.session_id core rider events session_id is not null: # check name
  metadata:
    id: 90bde4fa-148b-4f06-bd5f-f15b3d2ad759
    ownership_slack: #dispatch-service-dev
    tags: [rides, core-data, high-priority]
  query:
    type: dsl
    data_source_id: hive.core.rider_events
    filters:
      - session_id = null
  condition:
    type: fixed_threshold
    max: 0
  notifier_group:
    pagerduty_policy: dispatch-service
    email: dispatch-service@lyft.pagerduty.com Verity is primarily focused on defining and enforcing data quality checks rather than defining complete data schemas. Schema validation is not a novel concept. Several methods exist for defining event data schemas in event-based systems, such as JSON Schema, Protocol Buffers, Avro, or storage formats like Parquet. The optimal choice depends on your tech stack, usage, and specific requirements. While data schemas are valuable for defining the overall structure of data objects or table rows, they fall short in capturing more sophisticated validation checks specific to consumers, such as data distribution, business rules, SLAs, and thresholds. Data contracts go beyond schema validation, which focuses on identifying . I personally find that JSON Schema offers here a more suitable and readable option, effectively separating these structural and syntactical validation capabilities from serialization or storage concerns. syntactical errors However, to address and enhance data accuracy, having an effective means of defining data checks allows for defining the other facet of data contracts. semantic errors This is where Verity DSL comes into play. Beyond semantical validation, data contracts offer another crucial aspect that deserves attention. From a syntactical standpoint, validation checks remain consistent regardless of the consumer or producer involved. The set of validation rules is not tied to any specific consumer or producer and can be defined once and for all as a single schema. However, the Verity data contract DSL offers a finer granularity defining small independent rules, which is particularly well-suited for this context: the semantic meaning and usage of the data vary depending on the specific consumer. Additionally, not all consumers need to utilize all of the properties of an object. Their expectations differ. This doesn’t imply they are contradictory (which would certainly be an issue), but rather complementary and distinct ones. Therefore, allowing all consumers to establish unique rules that, when combined collaboratively, can provide a comprehensive understanding of the semantic significance of data quality is quite crucial. And it’s this aspect that particularly resonates with me. Bear with me, this may seem like a stretch, but from my perspective, it’s worth mentioning. :) collaborative Data exchange enables different teams (producers and consumers) to collaborate effectively. Establishing a shared understanding of these data exchanges is paramount, just like APIs in traditional software development. In microservice architectures, a known as consumer-driven contracts (CDC) has emerged, where consumers define the expected behavior of APIs provided by producers. Producers are responsible for verifying these contracts before releasing new versions. collaborative testing approach I think data contracts share a similar collaborative spirit. Even though data validation is performed on real data, rather than at release time, and does not block releases, it’s based on cooperation and encourages teamwork between data producers and consumers. I strongly believe that this collaborative approach is key to improving data quality and should be further integrated into the process. Well, I’m a big fan of drawing parallels… Notice actually that this collaborative aspect is something mentioned also as part of Lyft’s verity charter. The VerityUI provides a streamlined data discovery experience via the Verity Homepage. Our full-text search on the Check Definition Metadata lets users see all the checks currently being enforced and their Check Results. This has useful aggregations like owning team, table name, and tags. I’m not entirely clear on how data contract issues are shared between consumers and producers through the Verity platform UI, but I definitely recognize the significance of collaboration through the dashboards: The producer of a data product interface can confidently ensure that they’re not inadvertently causing downstream breakages that they hadn’t anticipated. The consumer of the interface can rest assured that their reliance on the interface is not and will not be compromised. While Verity is a remarkable tool for defining data quality checks, it’s unfortunately not open-source. Fortunately, there’s another open-source data quality framework called Soda Core that provides similar functionality. Soda Core is a free and open-source command-line tool and Python library that enables data engineers to test data quality. It utilizes user-defined input to generate SQL queries that run checks on datasets in a data source to find invalid, missing or unexpected data. When checks fail, they surface the data that you defined as “bad” in the check. https://medium.com/@seckindinc/soda-core-the-simplest-open-source-data-reliability-tool-f35ae5b576d9?source=post_page-----415baad6c95b--------------------------------&embedable=true During a scan of a dataset, Soda Core evaluates the predefined checks to identify invalid, missing, or unexpected data. Here’s the equivalent check using Soda.core syntax for the Verity DSL test that was previously defined. name: rider_events_session_id_check
source: hive
query: SELECT * FROM rider_events WHERE session_id IS NULL;
raise_alert: true
threshold: 10
action: slack
message: "There are more than 10 rows that are null for the 'session_id' property in the 'rider_events' table. Please investigate this issue." soda run --check checks/rider_events_session_id_check.yaml Soda Core is a powerful tool for ensuring the quality of your data. It can help you to identify and fix data problems early, before they can cause issues for your business. It’s worth noting that Soda Core can also facilitate data quality checks for streaming data by seamlessly integrating with Spark DataFrames. While Verity’s data quality checks for Hive are applied to static datasets, the checks for streaming data need to be more lightweight and efficient. Data would typically be processed in small batches of events, with very low latency, making them suitable for real-time checks and specific use cases like anomaly detection. Finally, let’s mention that there are other data validation tools available, such as DeeQu, Great Expectations, … As we wrap up, I hope you have a clearer understanding of the steps you can take to enhance your data quality journey. We’ve seen two distinct approaches to data quality improvement, each with its own strengths and methodologies. One focuses on increasing visibility and observability, motivating data producers to raise the quality bar. The other prioritizes elevating the quality standard through a testing and validation-first approach. Both are complementary. Verity is not merely a domain-specific language (DSL) for defining data checks; it’s a centralized platform that empowers data practitioners to collaborate effectively. This platform helps producers and consumers align on data quality expectations, including format, structure, and accuracy. Verity’s data contract management capabilities could (is?) further enhanced by integrating with a broader set of features, such as metadata management and discovery, to address more sophisticated data quality needs. Similar to Airbnb’s DQ score, Lyft’s Verity fosters a collaborative feedback loop between data producers and consumers. By incentivizing and empowering each team to take ownership of data quality, Verity cultivates a supportive environment where data quality continuously improves. , Found this article useful? Follow me on Linkedin , and ! Please 👏 this article to share it! Hackernoon Medium Also published . here

Stories, News & Mentions about LYFT

Lyft’s Stories on HackerNoon

Lyft's News & Mentions