This is the 3rd post in a 4 part series. You can find the intro post here and the second post here.
As a refresher — we’re using the metaphor of the “black box” from Andy Grove’s High Output Management to help us think about scaling.
The black box is comprised of 3 parts — the input, the machine itself, and the output. If we translate this to how to think about your startup, we can say:
- Input → People
- Machine → Organizational Structure
- Output → Results
In this post, we’ll focus on the 2nd part — the machine or organizational structure. For earlier posts, you find the intro here and the second one on people here.
Once you have good people, you need to make sure that they’re producing the results your business needs. This means crafting an environment where people can do their best work. What you’re looking to do is create a machine that takes input and consistently produces the desired output.
With companies like Zappos experimenting with holacracy, it’s easy for startups to think that they should just hire smart people and let them run wild. However, the dream of non-management doesn’t always work out and at any reasonable scale, you’ll likely need to start thinking about implementing actual structure & management.
So to build the machine you have to first design it, establish how it works, and then ensure it continues to work well. For startups, this translates to organizational design, defining your culture/values, and then maintaining efficiency.
Organizational Design
As with any machine, the first thing you have to do is design it. This means defining the desired output and then working backwards to figure out how to consistently achieve it. A common way to draw & present this is through an organizational chart.
One of the most important reasons to create an org chart is transparency. Not transparency in the classical sense (being open/honest, etc)—but more about making sure that it’s clear to everybody on the team how work flows through the org and gets done.
Without a clear picture of this, your employees won’t be able to see how their work contributes to the end result. It’s then easy for teams to get silo-ed and focus on just their own piece instead of the global output. Why worry about the end result when you’re just throwing shit over the wall & into the black hole?
However, if everyone can see how work flows through the org, they’ll be able to see how their piece fits into the bigger picture. More importantly, it takes the focus away from the individual and places it with the collective output. It puts the responsibility of the group output on the whole company. And when the pipes of the company are transparent, it’s easier to spot inefficiencies and make things better.
The 1st org chart (via Wikipedia)
The org chart originates from a man named named Daniel McCallum. A self taught engineer, he became the General Superintendent of the New York & Erie Railroad in the late 1800's. It was a massive org with 5000+ employees and several railroad that spanned over 2300 miles (another example from Harrison Metal’s excellent General Management course).
Back then, running a large railroad operation was an unprofitable endeavor. The number of people, the scale and complexity made it almost impossible to manage. As McCallum notes:
“A superintendent of a road 50 miles in length can give its business his professional attention and may be constantly on the line engaged in the direction of its details; each person is personally known to him, and all questions in relation to its business are at once presented and acted upon.
In the government of 500 miles in length a very different state exists. Any system which might be applicable to the business and extent of a short road would be found entirely inadequate to the wants of a long one.”
So McCallum created the first org chart to detail out how everything would work together. The intent was to create transparency & accountability across the whole company. And while you likely don’t run a large scale railroad, these tools to create organizational transparency still work.
Culture & Values
Once you have your organizational structure defined, you have to make sure that people understand how things get done within this structure. This means explicitly defining your culture & core values. It can seem silly at first, but good core values separate great companies from good companies.
Why [values] matter is that they become the first principles you go back to when you make decisions. It becomes a way to align people […] and it also gives you a list with which you should be able to figure out what to do and what not to do.
Enron’s Core Values, displayed in their lobby.
It’s easy to make hollow core values that nobody cares about (see Enron). But in order to create core values that people actually believe in and guide the company, you have to make sure that you’re not just saying these things, but you’re actually living them.
Going back to Coupa, we had 3 values — Ensure Customer Success, Focus on Results, and Strive for Excellence. While they can sound a bit corporate, they were actually quite helpful. They were deeply ingrained into everything that we did — praise, promotions, bonuses, etc. Everything was always presented within the context of our core values and it helped us prioritize, decide what to do (or not to do), handle difficult situations, etc. And as mentioned in the last post, a great place to start living your values is during onboarding.
There are plenty of posts and examples of good core values, so I won’t get into that here. But remember that it’s critical to define these early, because your culture starts with your first hire. And while in the beginning it’s easy to maintain cultural cohesion, it will get diluted if you don’t manage it. So use core values as a framework to maintain your culture as you scale.
Maintaining Efficiency
In addition to designing the machine & making sure it’s working, you want to make sure it stays efficient as the org continues to grow. There are 2 aspects of this — centralization / decentralization & eliminating roadblocks.
Centralization & Decentralization
When scaling, your input/org/goals keep getting bigger. As this is happening, you want to make sure that each individual part of the machine is still performing at the highest possible level (producing the max output). It can be easy for people to either get stuck in process or get overburdened. You can combat this through centralizing or decentralizing certain areas.
Centralization works when there is heavy overlap and consistency across a shared function. For example — as Coupa grew, our sales team was getting clogged up in process and in turn, had less time to sell. To move faster, we pulled out a few common sales functions and create a centralized team to deal with this — sales ops. This allowed our sales team to focus on what they did best and let our sales ops team handle the processing.
Decentralization is the inverse. If the areas are distinct enough, you want to break them up so coordination doesn’t get in the way. Again at Coupa, our engineering team started out as a centralized team. Every few weeks we would reassess where we were and self organize around important areas. But as we got bigger, we slowed down — things took longer and we shipped less. So we decentralized the team & divided them up by product. This allowed each unit to go deeper into their area, ship faster, and have more ownership.
While we didn’t invent either of the above, they’re both examples of how centralizing / decentralizing be used to maintain efficiency & output.
Removing Roadblocks
As any company grows, mistakes are inevitably made. It’s human nature to want to prevent them from happening again. One of the common tools people use to prevent mistakes from happening again is by creating rules.
In many cases, you need rules. But it’s easy to end up creating too many rules (both documented & undocumented ones). And when there are too many rules, it can feel like there’s too much red tape to actually get things done.
So your job is to find & eliminate unnecessary rules. Instead of rules, many things can be solved with principles instead. For example at AirBnB, they had a rule where all expenses required pre-approval. To cut through it, they killed the rule and instead lived by ‘use your best judgement’. Not only did this save a lot of time, but spend didn’t increase. Win-win-win.
The USS Santa Fe (via Wikipedia)
Another great example is from David Marquet, former commander of the USS Santa Fe. He took the ship from “worst to first” and achieved the highest retention and operational standings in the navy.
In his book Turn The Ship Around, David recounts a situation when a petty officer broke an important rule — when operating the submarine, you’re never to touch a machine with a red tag on it. The details are a bit esoteric, but suffice it to say—when you’re submerged deep in the ocean carrying nuclear weapons, it’s important to make sure your crew isn’t pulling on handles covered in red tags.
When the officer broke this rule, the team did a post mortem. While brainstorming how to prevent this from happening again, their ideas naturally gravitated towards new rules — new required trainings, more supervisors / approvals, etc.
However, the team decided to try and remedy this through principle instead. They implemented what they called ‘Deliberate Action’ — so before you did anything, you would physically say out loud what you were about to do. By doing this, they hoped to combat the mindless ‘autopilot’ mode people sometimes fell victim to when performing mindless tasks.
It wasn’t a rule and there was no punishment if you didn’t do it. But through this principle, the team achieved one of the highest performance ratings and per David, was the “single most powerful mechanism for reducing mistakes and making Santa Fe operationally excellent.”
If the commander of a large submarine housing 100+ people underwater while conducting stealth missions with nuclear weapons can build a great org without rules — then your startup can too.
To sum up — in addition to hiring good people, you have to make sure that you’re building a place where people can do their best work.
So design a transparent system, define how it works, and maintain efficiency through centralization / decentralization as well as removing roadblocks.
In the next (and last) post, we’ll sum up how to think about the results your organization is producing.
How To Scale Your Startup Part 1: Intro
How To Scale Your Startup Part 2: People
How To Scale Your Startup Part 4: Results
If you liked this post, click on the heart below and follow me on Twitter (@chriseyin). Any feedback / thoughts, comment or email me anytime.