The Bet That Should Have Worked
In March 2009, Foursquare launched at South by Southwest — the same conference where Twitter had its breakout two years earlier. The timing felt significant. Smartphones had just become mainstream. GPS was embedded in every new handset. And Dennis Crowley, who had already built and sold a location-aware social network (Dodgeball, acquired by Google in 2005 and then neglected to death), understood better than almost anyone that the phone-in-pocket represented a genuinely new input signal for consumer products.
The check-in mechanic was clever and novel. Arrive at a coffee shop, tap a button, announce your presence, earn a badge, accumulate "mayorships" for repeat visits. The gamification layer was lightweight but sticky enough that by 2013 Foursquare had 50 million registered users and check-in volume in the hundreds of millions. This was not a fluke. The team had built something people actually used, on a signal — real-time location — that no consumer app had successfully commercialized before.
The problem, which was visible in the data by 2011 but not publicly acknowledged until later, was that check-in behavior and monetizable intent were not the same thing. People checked in. They did not, in meaningful numbers, use Foursquare to decide where to go next. The product had successfully gamified presence without successfully replacing the "where should we eat tonight" search that was Foursquare's actual commercial opportunity. The distinction sounds minor. It cost the company several years of strategic clarity.
The Decision — Platform or Consumer?
By 2012, the company was sitting on an asset that few people fully appreciated at the time: a cleaned, disambiguated, continuously-updated graph of venues, categories, and foot-traffic patterns that no competitor could replicate from scratch. The raw material had come from years of user check-ins — millions of people voluntarily tagging locations, correcting venue names, adding new places, and building a ground-truth dataset of where people actually go.
Twitter was using Foursquare's location data. Snapchat's geolocation features were partly powered by it. Apple, Microsoft, and Samsung had all, at various points, either licensed or discussed licensing the Places API. The B2B platform play was not hypothetical — there were already paying customers for the underlying data layer.
The decision the leadership team faced in 2012-2013 was whether to pursue the platform business aggressively while the consumer app was still growing, or to stay the course on the consumer product and hope engagement deepened into monetizable behavior. They chose a middle path that satisfied neither option. The B2B data business was treated as a revenue supplement rather than the primary strategic bet. The consumer app received continued product investment in discovery features — Explore, local recommendations, personalized tips — that were genuinely good but arrived into a market where Yelp had been there for five years and Google Maps was improving its local recommendations at a pace Foursquare could not match.
The 2014 pivot — splitting the app into Foursquare (discovery) and Swarm (check-ins) — was the explicit acknowledgment that the original product had been trying to serve two incompatible use cases simultaneously. Swarm was for the loyal check-in users who cared about game mechanics and social presence. Foursquare was supposed to become the discovery experience that competed with Yelp and Google. The problem was that the split diluted both products without resolving the underlying question of what the company's primary bet was.
The meeting where the platform question was actually settled probably happened quietly, in late 2014, when the consumer Foursquare app's growth stalled post-split and the Places API revenue line continued to grow without a corresponding increase in investment. Not a dramatic pivot announcement — just a resource allocation signal that the organization gradually read correctly.
What Worked, What Failed
The data platform worked. By 2019, Foursquare's enterprise business — Places API, Attribution analytics, Audience targeting — was generating real revenue from marquee clients. The location intelligence product (which eventually merged with Factual in 2020 to create a scaled data company) was genuinely valuable and not easily replicated. The moat was real; it had just taken ten years and a failed consumer product to recognize it.
What failed was the consumer discovery app in its second incarnation. The post-split Foursquare app was not bad — the recommendations were solid, the venue database was excellent — but it had no mechanism to acquire new users, no structural advantage over Google Maps for mainstream consumers, and no clear reason for a non-power-user to choose it over the apps that were already on their home screen. The company had the right data and the wrong distribution.
The gamification layer — the one thing Foursquare had invented that competitors hadn't — was largely stripped from the discovery app in the attempt to make it feel like a "serious" recommendation product. This decision is debatable. The check-in game had been the user acquisition engine; removing it to appeal to a broader audience left the discovery app without either its original loyal users (who went to Swarm) or the mainstream Yelp-replacer audience it was trying to attract. Consumer product pivots that simultaneously change the core mechanic and the target audience tend to end up serving neither.
The optimism bias that kept the consumer bet alive too long is worth naming explicitly. The check-in numbers were impressive, and impressive numbers are psychologically easy to conflate with product-market fit. The distinction that mattered — whether users were engaged with the product or engaged with the game — required a harder look at behavioral data than early-stage teams typically take when growth is heading in the right direction.
What a PM Should Take From This
The Foursquare case is one of the cleaner examples of a company that had two distinct businesses — a consumer engagement product and a B2B data platform — and spent five years treating the less valuable one as the primary bet because it was more exciting and more visible.
The lesson is not "B2B is better than B2C." It's about how to read the signal when your competitive advantage is accumulating in a place that isn't where you're looking. Foursquare's data moat was growing year over year through the consumer app, but the consumer app's core user behavior — check-ins — was never generating the monetizable signal (intent-driven discovery) that the product narrative required. If you find yourself in a similar position: separate the behavior users are actually doing from the behavior your product thesis requires, and be rigorous about whether those are converging or diverging.
The pivot architecture also offers a cautionary note. Splitting a product to serve two incompatible use cases is sometimes the right call, but it only works if you have a clear answer to "which half is the future of the company?" Foursquare launched two products without clearly answering that question internally, which meant both products received insufficient investment and neither reached escape velocity as a standalone. The split felt decisive from the outside; inside it was a deferral.
For any PM evaluating whether to double down on a consumer product or shift toward the platform layer underneath it: the question is not which one is bigger. It's which one can you win, at what cost, and how defensible is the position once you get there. Foursquare could win the data platform. It could not win consumer local discovery against Google with the resources it had. That calculation was available in 2012. Acting on it earlier would have cost less.