Guide · Subscription apps
Onboarding is part of your acquisition cost.
Every person who installs and never reaches the paywall was paid for. That makes the first ninety seconds of the app a media buying problem, even though nobody on the media buying side ever gets asked about it.
The arithmetic nobody runs
You pay for installs. You earn from subscribers. Everything in between is a multiplier on your acquisition cost, and onboarding owns the first and largest step of it.
Move install-to-paywall from 40% to 55% and your effective cost per trial drops by more than a quarter – without touching a bid, a keyword or a creative. There is rarely a media buying change available with that kind of leverage.
Find the activation moment first
Every app has one: the point at which a user notices the thing they downloaded it for. A training plan appearing. A first meditation finishing. A document actually converting.
You can find it in your own data rather than guessing. Take users who converted to paid, look at what they did in the first session, and compare against users who churned. The action that separates them most cleanly is your activation moment.
Everything in onboarding should then be judged by one question: does this get someone to that moment faster, or does it delay them?
Screens that earn their place
| Screen type | Earns its place when | Is friction when |
|---|---|---|
| Question about the user | The answer visibly changes what happens next | It only populates a database |
| Feature explanation | The feature is genuinely non-obvious | It describes what the next screen would have shown anyway |
| Permission request | Asked at the moment the user needs the thing it enables | Asked on screen two, before any context |
| Account creation | Required to deliver personalised value | Placed before the user knows if they want this |
| Social proof | The category has a credibility problem | The user already installed – they are past that objection |
The permission row is worth dwelling on. Asking for notifications before anyone has seen the product produces a low acceptance rate and burns the request permanently – on iOS you get one prompt. Asked after the activation moment, framed around something specific, acceptance is far higher and the user is agreeing to something they can picture.
The gap between the ad and the first screen
This is the part that connects onboarding to media buying, and it is almost never instrumented.
Somebody taps an ad about sleep. They open the app and land on a generic welcome carousel about mindfulness, calm and focus. Nothing is wrong exactly – but they now have to work out for themselves that this is the sleep thing, and a measurable share do not bother.
If your onboarding is remotely configurable, matching the first screen to the acquisition source is one of the cheapest wins left in mobile. Search traffic told you what they want; social traffic told you what caught their attention. Those deserve different openings.
Measure by source, not in aggregate
Step-level drop-off across all users tells you where people leave. Drop-off split by acquisition source tells you why, and the two have completely different fixes.
If one campaign's users drop at the same step as everyone else, the step is the problem and onboarding needs work. If one campaign's users drop earlier than everyone else, the campaign is promising something the app does not open with – and that is a creative fix, not a product one.
Questions people actually ask
- How many onboarding screens is too many?
- The question is not the count, it is whether each screen earns its place by either collecting something you will use or delivering something the user notices. Four screens that personalise the product beat two that explain it, and both beat six that neither collect nor deliver.
- Should onboarding ask questions before showing value?
- If the answers visibly change what happens next, yes – being asked about your goal and then seeing a plan built around it is itself a demonstration. If the answers only populate a database, the questions are friction wearing a personalisation costume.
- Where does the paywall belong?
- Usually just after the first moment the user notices something worked. Before that and you are charging for a promise; long after it and you are giving away the value you meant to sell.
- Does the ad affect onboarding?
- Enormously, and almost nobody instruments it. Someone who tapped an ad about sleep expects the app to be about sleep. If they land in a generic welcome flow they do the translating themselves, and a measurable share stop translating and leave.
- How do you measure onboarding properly?
- Step-level drop-off by acquisition source, not in aggregate. The aggregate tells you where people leave; the split by source tells you whether it is the onboarding or the traffic – and those have completely different fixes.
- What is the fastest thing to test?
- Removing a screen. It needs no design work, no copy, no build beyond a config change if your onboarding is remotely configurable, and it produces a clean answer in days rather than weeks.
See what a few points of onboarding is worth.
The calculator turns install-to-trial rate into the maximum you can afford to bid. Move the number and watch the ceiling move with it.