Opinionated Defaults for Onboarding
Encode what you've learned works into product defaults — make the right choice easy and the wrong choice hard, without removing choice.
- Difficulty
- Advanced
- Time to result
- ~months to results
- Steps
- 3
- Confidence
- 92%
A product design pattern for onboarding: instead of leaving new users to make every setup decision cold, you bake your hard-won knowledge of what works into the defaults. You make it hard (add friction) to do the wrong thing and easy to do the right thing, but you never eliminate choice. The pattern emerges from first learning what works via human intervention, then productizing that learning as guardrails and recommendations.
Origin
Adam Fishman's term from his time leading growth at Patreon; he notes Airbnb independently arrived at the same idea and called it 'smart defaults.' At Patreon it was one of their product principles, informed by a data-science team analyzing hundreds of thousands of creators.
Core principles
- 01Onboarding is the only part of the product 100% of users touch, and the moment your brand promise must match your product delivery
- 02Users are most motivated during onboarding, so it's the right time to add friction and learn a lot about them
- 03Learn what works from human intervention first, then productize it so the non-scalable becomes scalable
- 04Guide behavior with defaults and friction, but preserve the user's freedom to override
How to run it
- 1
Identify high-potential users during onboarding
Instrument onboarding to learn who the user is — at Patreon, asking creators to connect accounts (YouTube, Spotify, Instagram, etc.), pulling in the data, and identifying who had large and heavily engaged followings per channel.
Pro tip Work with a data-science team to decide what actually predicts success per channel rather than inventing the criteria yourself.
Watch out Audience size alone is misleading — a million subscribers means little if they don't watch, comment, and share; measure engagement, not just reach.
- 2
Apply human intervention to the highest-potential users
Siphon high-propensity users off the self-guided flow and land them 'in the lap of a human being' who engages them, gets them excited, and coaches them on the best actions to take.
Pro tip Do this manually first precisely because it doesn't scale — it's how you discover what the right advice actually is.
- 3
Productize the human advice as opinionated defaults
Turn the advice the humans gave into in-product defaults and guardrails: pre-set the recommended configuration, add friction to changing it, and surface recommendations inline (e.g. Patreon nudging creators toward a 3-tier pricing model and a $3-5 entry point rather than a single tier or 40 tiers).
Pro tip Keep the override possible — a creator could still set one tier or a $1 tier, but the added friction steered most toward what worked.
Watch out Don't remove choice entirely; the pattern is making the wrong thing hard, not impossible.
In the wild
Patreon had learned across hundreds of thousands of creators that a three-to-five-tier structure with a $3-5 entry point worked best. They set those as defaults and added friction when creators tried to set up a single tier, 40 tiers, or a $1 lowest tier — while still allowing it. The upstream human-intervention program (connecting high-potential creators with a person at the right time) had already lifted first- and second-month creator revenue by 25%, and second-month revenue was a key input into lifetime value.
→ A 25% lift in early creator revenue that flowed through to a roughly equivalent lift in overall LTV on the platform, later made scalable by productizing the human advice into defaults.
Airbnb made instant-book the default for new hosts, which required host calendars to be accurate from the start. They worked to detect whether a host was a professional who knew what they were doing or a mom-and-pop just trying it out, and defaulted the calendar accordingly — fully blocked, fully open, or in between.
→ Reduced the risk of a new host getting an unwanted booking and quitting, converting a conversion moment into a retention safeguard.
Common mistakes
Eliminating user choice instead of adding friction
Removing options outright breaks the pattern and frustrates users who have legitimate reasons to differ; the mechanism is to make the wrong choice harder (more friction) while still permitting it.
Setting defaults from opinion rather than learned evidence
Opinionated defaults only work when the opinion is earned from data across many users and human-intervention learning; guessing at defaults imposes friction on the wrong behaviors.
Is it for you?
Best for
Growth and product teams optimizing onboarding for a two-sided or creator/host marketplace where setup decisions strongly predict success.
Not ideal for
Products where there is no learned 'right way' to configure things, or where user configurations are genuinely idiosyncratic and no default generalizes.
From the transcript
“one of our product principles which became opinionated defaults which is basically making it hard to do the wrong thing when you're setting up your…”
“they could still set up a single tier when we knew that a three-tiered pricing model actually would work better for them or they could…”
“what is not scalable when you're using people becomes very scalable if you can replicate components of that with technology”
“the idea of connecting someone with a human being the right person at the right time would improve the first month of Revenue or the…”
“it's one thing if you make crazy cat videos on YouTube and you have a million subscribers it's another thing it's every time you publish…”
From the episode
How to build a high-performing growth team
Adam Fishman (Patreon, Lyft, Imperfect Foods)