Ten thousand ads later, every screen is an ad

An ad gets roughly two seconds to earn two more. Nobody owes it anything. If the first frame doesn’t land a payoff, the thumb keeps moving and no amount of craft below the fold matters. When you’ve watched thousands of variants live or die on that rule, you stop believing anyone reads anything, and you start designing for the glance.

Product teams like to believe their users are different: invested, patient, willing to explore. They aren’t. Nielsen Norman Group has been publishing the same finding for two decades: people don’t read, they scan, and they scan with an exit in mind. Every screen in your product is competing with the home button. The screens that survive are the ones that answer “why am I here and what do I get” before the user finishes forming the question.

Duolingo’s splash screen is the cleanest example of ad discipline applied to a product surface I know. One mascot, one line (“the free, fun and effective way to learn a language”), two buttons. That line survived years of testing because it does three claims in nine words and then gets out of the way. Most fintech splash screens would kill for that economy.

We had a crude test for this in my banner years: flash the ad at someone for two seconds, cover it, ask what was on offer. If they couldn’t answer, the ad went back, no debate. I still run the same drill on product screens. Put the mockup up, count two, take it down, ask the room what the screen wanted from them. It’s slightly cruel and it settles arguments faster than any heuristic review, because there’s nowhere to hide: either the glance carried the message or it didn’t.

The streak screen is the best performing ad I’ve never been able to beat. One image, one number, one line of copy, one button. It isn’t reporting your progress; it’s selling tomorrow’s session while the endorphins from today’s are still warm. Single message, emotional payload, clear call to action: that’s an ad, and a disciplined one.

Headspace is interesting because it shows both instincts in one flow. The trial screen is honest ad copy: it tells you the price, the date you’ll be charged, and that a reminder is coming before it happens. That removes the exact fear that stops trial signups, and I’d bet it converts better than hiding the terms. Then on the welcome screen the terms checkbox arrives pre-ticked, an old growth-hack reflex the rest of the flow doesn’t need. The honest screen sells better than the sneaky one.

Once you see the pattern, the whole retention layer of modern apps reads as a media plan. Duolingo doesn’t only show your streak; it runs a “What’s a streak?” button under it, a product explaining its own ad format to a new user. The company has written openly about how the streak builds the habit, and the growth story behind it is dissected in Lenny’s Newsletter. It’s the most successful always-on campaign in consumer software, and it has zero media spend.

Headspace runs the same play with a shelf of badges where seven of eight are grey. Empty badges are ads for a future version of you; the greyed-out state is the creative. Brilliant borrows the honest-trial timeline that Headspace popularised, day by day, charge date included, which tells you the format is winning A/B tests across the industry. Good formats spread exactly like good ad formats do: someone proves it converts, everyone ships their version within a year.

And then there’s the push notification, the purest ad unit in software: a dozen words on a lock screen, competing directly with messages from people you love. Duolingo understood this earlier than anyone. The owl’s passive-aggressive reminders became memes, the memes became free reach, and underneath the joke the mechanics are dead serious: the notification never sells the lesson, it sells the streak you’re about to lose. Loss beats gain in copy tests as reliably as it does in behavioural economics papers, and nobody spends that insight more shamelessly, or more effectively, than a green cartoon bird.

None of this is cynical by default. A streak is a promise kept daily, and an honest trial timeline is the least manipulative paywall ever designed. The cynicism only arrives when the ad writes a cheque the product can’t cash. That’s as true inside the app as it was in my banner years.