4 min read
Optimizing for day-one conversion is the wrong default for subscription products
Flo gets 40% of its subscriptions more than six months after install—intentionally. That single data point should make most subscription teams question what their conversion funnel is actually optimizing for.
By intermac
Most subscription products treat a low day-one conversion rate as a bug. Flo treats it as a feature. According to Mike Charniatsou, who works on growth and web onboarding at Flo, 40% of all Flo subscriptions happen more than six months after install. That is not a leaky funnel—it is the product philosophy made concrete.
Mike's line that stuck with us: "Your paywall conversion is not your monetization strategy." That's worth sitting with. A growth team can run a clean A/B test, see a 10% lift on the paywall, ship it, and lose money. The BoldVoice example from the same conversation illustrates this exactly: removing the X button from a paywall lifted conversion and spiked refunds until the win became a drain. A paywall metric and a business metric are not the same thing, and conflating them is an easy mistake to make when you're under pressure to show weekly wins.
The lifecycle that most funnels ignore
The case for delayed conversion only holds if your product has genuine reasons to re-engage a user months later—if circumstances in their life change what they need from you. Flo has that: cycle tracking is free, but pregnancy support, fertility tools, and perimenopause guidance are distinct enough that a user who didn't need them in January may absolutely need them in September. The monetization follows the need.
Most subscription products don't have that kind of natural re-engagement arc. If your app solves one problem and solves it well, waiting six months for conversion isn't a strategy—it's just churn. Mike is honest about this: a younger app under acquisition pressure can't afford the patience Flo has built over years. So this isn't universal advice. But the underlying principle—that conversion is a lifecycle, not a single event—applies to almost everyone.
Evidence over explanation
The experiment story in the article is the most practically useful part. Flo's team added an onboarding step to identify a user's perimenopause phase. It dropped conversion by 10%. They ran it again. Same drop. Only then did they look at the data, and found that 75% of users couldn't be assigned any phase at all—meaning the screen they saw after clicking an ad promising to identify their phase told them: we couldn't do that, carry on.
The team's first instinct was to explain the failure rather than examine it. Mike's rule that follows: you iterate when you have new evidence, not a new explanation. Explanations are unlimited. Chasing them is a "vicious cycle of experimenting" that never lands anywhere. The fix—rewriting the question with the medical team and routing unclassified users past the dead end—came back with a 23% win on the third attempt, confirmed by a fourth blackout test.
That discipline is harder than it sounds when you have a strong prior hypothesis and stakeholders waiting for results. But it's the only way to know whether you're learning or rationalizing.
What to take from this
If you build a subscription product, two things are worth checking. First: sum up your reported experiment wins from the past year and ask whether your business actually grew by that percentage. If it didn't, the gap is wins that were never business wins. Second: when a test fails, ask what evidence would change your next iteration before you start building the next iteration. If the answer is "none in particular," you're chasing an explanation.
The broader point is simpler. Day-one conversion is a proxy. The real question is whether your product has a reason to still matter to someone in six months.