Playable Ads — A new ad format for Meta

Proving it could work

It started from zero: just me and one engineering partner, running a baseline test on Facebook Feed. I scoped it end to end. The test ran the old single-CTA format, and in most campaigns it didn't perform as well as video on installs. Deeper in the funnel it flipped: on purchases and retention, playables beat video. People who tried the demo first installed with more conviction, and stuck around longer.

Initial experiment on Facebook Feed.

That was the evidence, but I didn't pitch it with a deck. I built a quick, working prototype. Rough, but functional enough to show in context. A ten-second demo communicated the concept faster than any document could. Leadership got it immediately, and the conversation went from "explain this format" to "when can we test it."

Two user groups, one ad unit

The same test explained the install gap: users with high intent just wanted to go straight to the download, and a single "Try demo" CTA stood in their way. So the design had to serve both intents at once: "Install now" for users ready to convert, "Try demo" for the ones who wanted to play first. No model for showing both existed on Instagram, so I designed five concepts for how it could work on Reels.

I fought for the mid-scene overlay. Getting a new CTA pattern onto Reels meant a rigorous review, so I anchored the conversation with aggressive concepts first (playables on impression, full-surface takeovers), then presented mid-scene as the measured alternative. It worked.

Different variants of the dual CTA in mid-scene.

Before and after. Left: the single-CTA control that removed the install path. Right: the dual-CTA design we took live, keeping both options in one ad unit.

Shipping meant winning a separate design review for every surface across both apps, working with each format owner in turn. iOS and Android each needed their own solution.

The end-to-end flow, from Reel to install.

What shipped is the mid-scene concept. The ad starts as normal video, then mid-play the unit surfaces both CTAs. Tapping "Try demo" opens the playable, an HTML experience inside Meta's in-app browser, with a persistent install CTA anchored at the bottom. On iOS, Apple's SKOverlay (a native component that shows basic app metadata) lets users install without leaving the ad.

Designing Android's SKOverlay from scratch

Android has no equivalent of Apple's SKOverlay, so I designed one from scratch. Same goal (app metadata, ratings, one-tap install), built for a platform without the convention.

Apple's native install overlay on iOS, and the custom equivalent I designed for Android.

What I cut

I designed multiple iterations of playable ads on impression: letting users interact with the demo directly on the surface, no post-click transition. It was the most ambitious concept and I pushed hard for it. It got rejected. I agreed with the call.

Playables on impression. Rejected.

Users don't expect an ad to behave like an app. Dropping a game straight into a Reel breaks how people read the surface. That post-click tap matters: it's the moment a user chooses to enter the demo. Take the choice away and the format just feels invasive.

Playables are now a standing part of Meta's app-install business, live at full scale across both apps.

Advertisers adopted it immediately. They were already producing playable assets for AppLovin, so running them on Meta took no extra creative production, and it opened a path beyond gaming into e-commerce and other verticals, anywhere letting people try first makes the download feel less of a gamble.