3 in 4 Growth Bets Fail: The Test Founders Skip

When Tallo asked me to look at two possible new markets, I came back with a yes on one and a no on the other. The no was the more valuable answer. The research showed the second market didn't have enough customers in it to be worth building for, and saying so before anyone wrote a line of code saved the team months they would otherwise have spent finding that out the expensive way. I wrote that up as a case study, and it's the kind of answer Peter Sellis was arguing for last month.

Sellis was the first product manager at Snapchat and later Discord's head of product. On Lenny's Podcast in September, he argued that growth almost always comes from the core product, not from chasing new users or new markets. The full episode is behind Lenny's paywall, so I won't quote him directly, but the argument is simple to state and harder to act on: before you go looking for somebody new to serve, make sure you've gotten everything you can out of the people and the product you already have.

That cuts against the instinct most founders have when growth flattens, which is to find the next market. So it's worth looking at what the evidence says, and at the test I'd run before any expansion bet.

Snap and Discord both went back to the core

Snap is the cleaner public example. In the second quarter of 2018, Snapchat's daily active users fell for the first time in the company's history, from 191 million to 188 million, according to Snap's own results. CEO Evan Spiegel blamed the drop on "a slightly lower frequency of use among our user base due to the disruption caused by our redesign." Spiegel also credited the redesign with an 8% increase in retention among users over 35, so it was winning over people Snapchat hadn't reached before while the people who already loved the app used it a little less.

In Sellis's telling, what came next wasn't a hunt for a new audience. The team went back to the experience of the users it already had, with a lot of attention on how the app performed on Android, and that's where the next stretch of growth came from.

Discord's version is more recent. In May 2024, its CEO wrote that the company had recognized the need to narrow its focus "from broadly being a community-centric chat app to being a place that helps people deepen their friendships around games and shared interests." That was a reversal. During the pandemic, Discord had pushed to become a general-purpose place to hang out, and a month before the announcement its CEO had pointed out that 95% of its users play games. Sellis credits that refocus on gamers, an audience Discord supposedly already had covered, with some of its fastest growth since the pandemic.

I can't verify the growth figures he cites from the outside, and I'd rather not pretend to. The direction of both stories is what matters: two companies with real audiences found more growth by serving those audiences better than by reaching for new ones.

Most moves away from the core don't pay off

None of this is new to anyone who's read Bain's work on growth. Chris Zook's research there found that only about one in four "adjacency moves" – a new geography, a new customer segment, a new channel, a new product – generated profitable growth. That research is from the early 2000s, so I'd hold the exact number loosely, but the shape of it holds up in my experience: the further a bet sits from what a company is already good at, the worse its odds.

The Bain work also gives a useful way to think about distance. Count how many things change at once. A new market served by the same product, through the same channel, solving the same problem, is one step from the core. A new product for a new audience through a new channel is three or four steps out, and every step is a place for the bet to break.

The test before "which market next"

Before anyone sizes a new market, I'd want answers to three questions.

Are your current customers getting value as often as they could? Sellis frames this for consumer products as how often people come back, because a daily user is worth far more than an occasional one. The B2B version is depth: how much of your product the average account actually uses, how many people inside it use it, and whether the ones who leave are leaving because something better came along or because they never got the value they bought. If those numbers have room to move, that's growth you can reach with a team that already knows the customer.

Is anyone already using you for the new thing? Sellis's signal for a real expansion is users going out of their way to use a product for something it wasn't built for. If the new market is real, some of your current customers are probably already doing a clumsy version of it. If nobody is, that doesn't kill the idea, but it means the demand is a hypothesis you have to go test, not an observation you can build on.

Is this real demand, or the comfortable alternative to a harder problem? This is the one I'd push hardest on. Organizations reach for something new to build because building is the path of least resistance. Product and engineering will always say yes to a new product, and it feels like progress. Meanwhile the actual problem might be that the core product has stopped getting better for the people who already pay for it, which is a much less exciting conversation to have in a board meeting. A new market can be a way of not having that conversation.

When expansion is the core, carried somewhere new

The point isn't that companies should never expand. Some of the best growth I've been part of was expansion, and it worked because it was the core product carried to a new audience, with the demand visible before we built anything.

At CoinDesk, I built the machine translation system that took our journalism into new languages, and every language we launched grew 70–120% a week. It was the same product, the same reporting, for readers who'd been underserved in their own language. We picked the order of languages from data about where people were holding and trading cryptocurrency, so each launch went where the demand already was.

Tallo's yes worked the same way. Career starters were a new audience, but Tallo's core strength – helping young people find work – was exactly what they needed. The second market failed a simpler test: there weren't enough people in it.

Where B2B founders should adjust the advice

Sellis is talking about consumer social products, and some of it doesn't translate cleanly. Daily use is the right measure for a messaging app and the wrong one for software a finance team opens at month-end. In B2B, growth from the core often looks like expansion inside accounts you already have: more seats, more teams, more of the product in use. That's still core growth, and it's usually cheaper than a new segment, because you already have the relationship and the proof.

What does translate is the order of operations. Find out what your current customers aren't getting from you yet before you go find new customers. A cheap yes leads to an expensive no, and "let's go after a new market" is one of the cheapest yeses in a planning meeting, because it postpones every hard question about the product you already have.

The next market will still be there after you've answered those questions about the one you're in, and you'll have a much better read on whether it deserves you.