Inside a playable ad, track six moments: the ad displayed, the first interaction, progress through the challenge, the pass or fail, the end card shown, and the CTA clicked. These in-ad events explain why a playable wins or loses; campaign metrics like IPM, CPI and D7 ROAS, which come from the network and your MMP, decide whether it stays live.
This page covers the in-ad layer. For the campaign layer and the full cut, iterate or scale method, start with how do you know a playable ad is working.
How are in-ad events different from campaign metrics?
They answer different questions, come from different systems, and should drive different decisions.
| Aspect | In-ad events | Campaign metrics |
|---|---|---|
| Examples | Displayed, first interaction, challenge passed or failed, end card shown, CTA clicked | Impressions, CTR, IPM, CVR, CPI, D1 and D7 retention, ROAS |
| Where they come from | Code inside the playable, reported by networks that support it | Network reporting and your MMP |
| Question answered | Why did players drop off or tap through? | Did the playable bring valuable players at a good price? |
| Decision | What to change in the next version | Cut, iterate or scale |
| Available on | Some networks (AppLovin, Mintegral, Unity, plus Luna Playable Insights for Luna builds) | Every network |
Teams we talk to often have plenty of playables live and still cannot say why one beats another. A creative lead at a large studio described their playable data as "all over the room." That is what happens when you only have the campaign layer.
Which events should every playable track?
Use AppLovin's event names as a common language, even on other networks. They map cleanly onto how a playable is built (see anatomy of a playable ad).
| Moment | What it tells you | AppLovin event name |
|---|---|---|
| Displayed | The playable actually rendered for the player | DISPLAYED (required); LOADING and LOADED optional, used as a pair |
| First interaction | The player understood the hook and started. Time from display to here is your time to engage | CHALLENGE_STARTED, fired on the first real move |
| Progress | How far players get before they stop | CHALLENGE_PASS_25, CHALLENGE_PASS_50, CHALLENGE_PASS_75 |
| Pass or fail | Which ending the player reached | CHALLENGE_SOLVED, CHALLENGE_FAILED, CHALLENGE_RETRY |
| End card shown | The player reached your pitch inside the playable | ENDCARD_SHOWN |
| CTA clicked | Interest turned into intent | CTA_CLICKED |
AppLovin's rules worth knowing, from its documentation:
Which metrics come out of those events?
| Metric | How to calculate it | Reference point |
|---|---|---|
| Engagement rate | First interaction ÷ displayed | Vendor claim (Sett, no method published): median drop-off in the first three seconds is 46.7% without guided interaction, 28.6% with a strong first-time experience |
| Time to first interaction | Seconds from displayed to first interaction | Google's goal for HTML5 ads: interaction within 3 or 4 seconds. Vendor claim (Sett): median 8.76 seconds |
| Drop-off point | Where the progress events stop | Compare versions against each other, not a market average |
| Completion rate | Solved or failed ÷ first interaction | Vendor claim (Sett): average play 35 to 40 seconds; successful playables 70 to 80 seconds |
| Win CTR vs fail CTR | CTA clicks after solved vs after failed | Test both endings; there is no universal winner |
| End card conversion | CTA clicked ÷ end card shown | Compare end card variations |
Treat the vendor figures as direction, not benchmarks. Your own previous versions are the reference that matters. More context on where these numbers come from is on our playable ad statistics page.
Know why it won, then build the next one. Genie Playable builds each version from the last, so the hook or ending that worked carries forward instead of getting lost in a spreadsheet. No developer needed.
Try GenieWhat does each network let you track?
| Network | In-playable analytics |
|---|---|
| AppLovin | Official API: window.ALPlayableAnalytics.trackEvent('EVENT'), wrapped in a check that the object exists. Predefined events only; each appears as its own report column |
| Mintegral | Supports up to five action (tracking) points, configured in an action.json file packaged with the playable, per third-party build tool documentation |
| Unity Ads | No network requests for content, but Unity can permit analytics calls that collect player interaction data, with no personal data such as user or device identifiers. Unity handles start, view and click attribution server-side |
| Unity Playworks (Luna) | Playable Insights collects standard events automatically, supports custom events, and breaks results down by playable, creative, ad network, OS and country. All data is anonymous |
| TikTok | No in-playable analytics API documented. When an ad group mixes video and playables, results are reported at ad group level only |
| Meta, Google Ads and others | Public specs do not document an in-playable analytics API. You get standard ad reporting |
Privacy applies everywhere. No user or device identifiers, no personal data. Track what the player did in the ad, not who they are. Packaging and size rules for each network are in our playable ad specs hub and the AppLovin specs page.
How do you connect in-ad events to ROAS?
Keep the two layers separate, then read them together. In-ad events explain behavior: whether the hook worked, whether the tutorial was clear, which ending drove the click. Network and MMP data tell you whether those players paid back.
The gap between the layers is real. UA consultant Matej Lancaric reviewed about 360 creatives for one game in an analysis published in August 2026 and found the creative that converted most efficiently was not the one that paid back best. Strong engagement does not guarantee strong D7.
A workable loop:
Our guide to iterating a winning playable covers what to change at each step.
What should UA teams ask for from whoever builds their playables?
Where does Genie Playable fit?
Genie Playable builds playables from your APK or HTML build plus the assets in Genie, from a format that is already working, or from scratch from a brief with Genie Agent. It exports optimized builds for Meta, AppLovin, Unity Ads, Liftoff, Google Ads, TikTok Ads, Pangle and Mintegral, with Moloco coming soon. Because every version comes from the one before it, what you learn from the data stays attached to the playable.
Genie Playable is not an analytics platform or an MMP, and it does not replace network reporting. If you need a dashboard that pulls creative performance across networks, that is a different tool. It is also not the right fit if your in-house team already ships instrumented variants as fast as you can test them.
Self-serve starts at $490 a month, or $249 a month for startups, with a free sign-up that covers your first two playables. The custom plan, usually for larger companies, is a full one-month pilot where the Genie team onboards you and helps you build the playables you want. The product is moving fast, and we will keep this page current.
Build the next version from what you learned. Track the six moments. Change one thing. Genie Playable builds the next version from your game, without a developer.
Try GenieFAQ
Sources
- 01AppLovin, Playable analytics integration
- 02Unity Grow, Playable specifications
- 03Unity Playworks (Luna), Playable Insights overview
- 04TikTok Ads Help, How to create Playable ads on TikTok Ads Manager
- 05Google Ads Help, Make better creative assets for App campaigns
- 06Needle Engine, Mintegral playable deployment (third-party build tool documentation)
- 07Matej Lancaric, How to find a winning creative, August 2026
- 08Sett, Playable ads complete guide (vendor claims)
