Onboarding as a Retention Lever
Treat onboarding as a habit-and-retention engine, not a conversion funnel — and accept lower conversion to weed out the wrong users.
- Difficulty
- Moderate
- Time to result
- ~months to results
- Steps
- 4
- Confidence
- 90%
A reframe of onboarding away from pure conversion optimization toward retention. Because onboarding drives early habit formation and churn concentrates in the earliest usage, improving activation can shift cohort retention curves outward by 10-20 points. Counterintuitively, doing onboarding right can lower conversion — you deliberately weed out poorly-fit users who would have churned anyway — and you validate the tradeoff with early proxy metrics rather than waiting 90 days.
Origin
Adam Fishman's synthesis from running onboarding projects at Lyft, Patreon, Wyzant, and Imperfect Foods. He connects it to a Collison-brothers idea he cites: user research informs your model of the customer, and that model informs what you build.
Core principles
- 01Onboarding drives habit formation, which drives retention; churn is most likely in the earliest usage of the product
- 02Getting users past the early churn hump is one of the most valuable things you can work on
- 03Higher conversion is not the goal — sometimes conversion should decrease as you filter out ill-fit users
- 04Don't redesign onboarding for its own sake; only revisit it when you have a genuinely new insight about users or your growth model
- 05Use proxy metrics because nobody can wait 90 days to learn whether a change worked
How to run it
- 1
Reframe the metric from conversion to retention
Recognize onboarding's biggest impact is on retaining users by connecting them to the product's value and forming a habit early, not on maximizing the number who complete the flow.
Pro tip Look at pairs of opposing push-pull metrics together (conversion vs retention) rather than optimizing one in isolation.
Watch out You can cut conversion so far that the retention gain doesn't offset it — that's the wrong direction, so watch the balance.
- 2
Accept and even welcome some conversion loss
If done right, fewer people may complete onboarding, but the ones who do are more qualified and more likely to be engaged, retained customers. Weeding out ill-fit users improves retention by construction.
Watch out This only holds if the people being filtered out were genuinely poor-fit; don't use it to excuse a broken flow that loses good users.
- 3
Validate with proxy metrics, not 90-day waits
Identify early leading indicators that predict long-term retention. At Patreon, velocity to first patron and first $100 processed strongly predicted eventual size and retention.
Pro tip Combine quantitative proxies with qualitative screening — sample onboarding cohorts, look at who the users are and whether they're doing the right behaviors, and run them through your model.
- 4
Revisit onboarding only on new insight
Fold net-new learnings about customers or your growth model back into onboarding, but resist micro-tweaking or full redesigns 'for redesign's sake,' which rarely move the needle once the flow is in a good foundational place.
Pro tip At scale (hundreds of people), consider a team whose roadmap owns the activation experience for a quarter or two a year, since onboarding is easy to build once and never revisit.
Watch out Coming into companies, Fishman repeatedly finds onboarding built years ago and never touched while retention silently suffers.
In the wild
Rather than waiting for 90-day retention, Patreon used how quickly a creator reached dollar thresholds after launch — velocity to first patron and first $100 processed — as proxies for eventual size and retention, backed by qualitative sampling of who was onboarding.
→ Enabled fast, confident evaluation of onboarding changes without long waits.
Fishman notes that at Airbnb a host could sign up, get booked on a day they didn't want to host, have a bad experience, and quit — making onboarding a retention problem, not just a conversion one. The fix was detecting host type and defaulting calendars appropriately.
→ Framed calendar accuracy at onboarding as protecting long-term host retention.
Common mistakes
Optimizing onboarding purely for conversion
Maximizing completion can pull in poorly-fit users who never form a habit and churn early, hurting the retention curve that actually drives the business.
Redesigning onboarding for redesign's sake
Micro-tweaking or full redesigns without a new hard-earned insight rarely move the needle once the flow is foundationally good; the outsized wins come only from net-new understanding of customers or the growth model.
Is it for you?
Best for
Growth/product teams whose retention curves dip in the first days of usage and who want to prioritize activation over top-of-funnel conversion.
Not ideal for
Pre-product-market-fit teams still searching for any conversion at all, or businesses where retention isn't the primary value driver.
From the transcript
“one of the things that onboarding really drives is habit formation with the product which leads to retention”
“I have seen companies shift their curve outward by 10 15 20 percentage points from making those changes especially when you consider that churn is…”
“I actually think if you're doing it right sometimes conversion should actually decrease a little bit so you might actually have fewer people getting all…”
“the answer is not wait 90 days to find out right like nobody has that kind of time and so what I tend to do…”
“I very strongly dislike the idea of redesigning something for redesigning's sake”
From the episode
How to build a high-performing growth team
Adam Fishman (Patreon, Lyft, Imperfect Foods)