by Jason Costa
I’m biased because I worked there for a time, but I’m constantly rooting for Twitter. It’s one of the quintessential communication tools in the world, and it is still *the* real-time information platform. I want to see that company win because the raw product is so good. That said, Twitter’s valuation is approaching single digits, and it’s increasingly unclear how the company is going to introduce non-linear product changes into the service, in a way that brings new users into the picture (and resurrects dormant ones).
Compare Twitter’s growth (~1.4%, average QoQ) above to Facebook’s, which has a user base more than five times the size of Twitter’s, yet is still growing at ~4%. Below is a reflection on 3 product misses that, in my opinion, dramatically impacted the trajectory of Twitter as a company.
The engagement stats on Tweets that contain photos, versus those that are purely text, were staggering. Undeniably, people engaged far more with Tweets that had media attached. But despite knowing this from internal data, photos continued to be relegated to 3rd party services integrated into the network such as TwitPic, YFrog, and Instagram. It wasn’t until late 2011 that Twitter made photos a 1st party, native function in the mobile product.
Speaking of Instagram (which back in mid-2010 was referred to as Burbn, before the photo project morphed), that service was growing at an insane rate. Inside of Twitter, one could see that the number of users connecting their Instagram account with Twitter had hockey stick growth, a proxy to Insta’s overall growth.
Twitter missed the opportunity to acquire Instagram, integrate that service into the product and company, and prioritize a vision for photos native to the mobile service. That could have placed Twitter in a very competitive position for owning the mobile photos space, especially considering that Facebook still subscribed to the religion of HTML5 back then. Photos were a major paradigm shift in social mobile, and missing a great native photo experience was a major loss.
In a bid to better differentiate from Facebook, Twitter would consistently position itself as the de facto “interest graph” in market, eschewing the “social graph” moniker that FB touted. It was a sound marketing strategy and gave users yet another reason to have a separate Twitter account in addition to their FB profile. Eventually this led to an identity crisis, and ultimately distracted the company from focusing on the user experience.
Twitter’s ability to import a user’s email contacts and mobile address book to drive graph density severely lagged Facebook. In 2010, Facebook even acquired an email scraper team out of Malaysia (Octazen) just to help FB map out the entire social graph. Twitter didn’t necessarily need to go that far, but it did need to do more up front to connect people with their real life social ties — even the initial Twitter NUX flow itself focused on topics rather than people.
As the company spent more time evaluating user data, it turned out that people who followed users with whom they had social ties were far more engaged and retained better than those who didn’t. Those push notifications you get about “person X from your address book just joined the service — say hello” — turns out they work! These help drive network density and impact retention. People like interacting with their friends, not just following celebrities, politicians, and athletes (these are obviously important nodes in the network, too).
Having a large “Following” graph and pruning down is a far better state to be in than having a graph with very little social value. Recognition of and connecting to real life friends were far too hard on Twitter. When someone walks into a party and immediately recognizes one or more of their friends, they’re going to feel a lot more comfortable about staying.
Early on, because of its real-time nature, Twitter adopted a deterministic, reverse chronological timeline that displayed the most recent Tweets at the very top. Over time the company held firm to this approach, and didn’t explore what an algorithmically sorted stream might look like. Facebook revealed that they were doing this as early as 2010 with EdgeRank. There were ample opportunities to bundle Tweets into interest and socially based stories. Instead, the home timeline was treated like a piece of infrastructure that just needed to be kept up and running.
There were many conversations about what simplified, interest-based Tweet digests might look like — and that would have gotten Twitter closer to television channels with things like clustered timelines (this was still long before live video streaming had picked up steam). The “In Case You Missed It” Tweet summaries as seen today have been a nice injection into the home timeline, and hopefully the company can continue to push the boundary further there. For instance, if a given user is a basketball fan (and Twitter has the data to know that), then show that user some of the incredible sports-specific conversations that unfold only on Twitter.
One interesting play back then would have been to acquire Prismatic and Dataminr, and then let those teams run timelines and notifications, respectively. Prismatic did an incredible job curating the most engaging content on Twitter into personalized digests, based on a specific user’s interests. DataMinr does impressive event detection by analyzing the Twitter firehose. If acquired, Dataminr could have compressed the time to interact with interest based events as they happened, by proactively notifying users. Prismatic could have enabled stories and surfaced related media around said events.
There would have been a complete story from notification to Tweet view to media consumption (and it would have made notifications far more unique as a product vector on mobile, compared to what’s out there today). But I digress with what could have been: the core issue was that there wasn’t enough boundary pushing on what the home timeline product could be. The home timeline should have been the primary surface area by which Twitter would help users connect to their interests, and discover what’s happening in the world around them.
It’s always easy to look back and pontificate — hindsight is 20/20. There have been a lot of posts written about “if Twitter did this or if Twitter did that,” but truthfully no one really knows what else “could” have happened had Twitter executed flawlessly back then. Facebook has only become an even bigger juggernaut and is giving their competitors all kinds of problems. But I am convinced that had Twitter nailed photos, made the experience more social, and expanded what the home timeline was — we’d all be having a very different conversation about Twitter today. Companies have a window in time when it comes to market opportunity, and a team has to execute ruthlessly to keep that window open.
Also important — you can’t reset strategy every six months. You must pick an approach and double down. This was an organizational issue more than a direct product issue though, so I’ll save digging in on this for another time. This post is purely meant to reflect on a set of strategic product circumstances that I personally believe impacted the company most. As others are building companies and products themselves, hopefully they can borrow from these learnings along the way….
Jason Costa is currently an EIR investing at GGV Capital. This post is part of an ongoing series aimed at exploring topics such as consumer product development, platform analysis, and strategy.