Onboarding Is the Product: Redesigning Aperture's First Run

Creative software has a brutal funnel truth: most trials are decided in the first session, and most first sessions end in quiet defeat. When we started working with Aperture on their first-run experience this summer, their data showed the classic shape. Strong signups, healthy day-one opens, and then a cliff, with the steepest drop in the first fifteen minutes of actual use. People were not churning from the product. They were churning from the blank canvas.

We just shipped the redesign, and the process rearranged how I think about onboarding in general, so here are the notes.

Time to first pride

The metric everyone uses is time to value, and it points teams at the wrong target. Value is the company's word. Nobody opens a creative tool seeking value; they open it hoping to make something they do not feel embarrassed by. So we adopted a blunter internal metric: time to first pride. How many minutes until a new user has produced something they would show another human?

That reframe changed the design immediately. The old onboarding was a tour: coach marks pointing at panels, a progress checklist, a sample project to poke at. Tours optimize for feature awareness, and feature awareness is worthless to someone who has made nothing. Users clicked through all seven steps and left anyway, informed and empty-handed.

The new first run is a guided act of making. You start inside a half-finished piece chosen to match the intent you declared at signup, and the first three actions you take are real ones, on real work, each producing a visible improvement. Within four minutes you have made something specific and yours. The panels get introduced only when the work requires them, which turns out to be the only time anyone can absorb an interface anyway.

What the testing kept telling us

We ran moderated sessions weekly for the whole engagement. Three findings that generalize beyond this product:

• Blank states are a courage problem, not a comprehension problem. Users understood the canvas fine. They froze because the first mark felt like a test of talent. Starting from something 80 percent done removes the test.
• Undo is an onboarding feature. Once we surfaced how safely everything could be reversed, exploration roughly doubled. People experiment exactly as much as they trust the floor.
• Skippers are not a failure case. Around a quarter of users bail out of any guided flow immediately, and they include the most committed segment: experts migrating from other tools. We built them a separate path, a keyboard-first quickstart, instead of grading them as churn risks.

Where AI actually helped

Aperture, like every creative tool this year, has generative features, and the temptation was to lead with them: type a prompt, receive a result, instant magic. Testing punctured that quickly. A generated result produces surprise but not pride, because the user cannot locate their own contribution in it. What worked was inversion: the AI prepares the situation, the human makes the moves. Generation as scaffolding rather than performance. The demo is less flashy and the week-four retention is meaningfully better, which tells you which one is the product and which one is marketing.

Early numbers since launch: activation on the guided path is up double digits, and the completion-to-subscription correlation is strong enough that the first-run team now sits with growth rather than with design. Which is really the conclusion. The first fifteen minutes are not a layer in front of the product. For a new user, they are the entire product, and they deserve the same craft budget as anything else you ship.