Playable Ad Funnel Metrics: From Impression to Install
A playable ad loses people at different stages for different reasons. This guide walks through each stage of the funnel, what a weak ratio usually means, and what to change first.
Key takeaways - A playable ad is a funnel with distinct stages: impression, first interaction, completion, end-card click, store page and install. Each stage fails for a different reason. - Read the funnel from the top down. A weak first tap points at the opening, a weak finish points at the mechanic or length, and a weak install step points at the end card, the store page or the audience. - Ratios between neighbouring stages are more useful than any single number, because they isolate one decision at a time. - Not every network reports every event. Know what you can measure before you plan a test around it. - Early-funnel metrics are diagnostics. Installs, and what those users do afterwards, remain the final judge.
Why think in a funnel at all
A single number such as install rate hides where the ad is losing people. If a playable underperforms, you have to guess whether the opening was confusing, the mechanic was dull, the end card was weak or the audience was wrong. Those problems need different fixes.
A funnel splits one big outcome into smaller questions. Each question has an owner: the first screen, the tutorial, the gameplay moment, the end card, the store page. When a ratio drops, you know which part of the creative to open.
This article covers the playable-specific funnel. For how the two headline metrics relate, see IPM vs CTR for playable ads.
The stages, in order
1. Impression
An impression means the ad was served. It does not always mean it was seen, and for a playable it does not always mean it loaded. Playables are interactive HTML, so they can fail in ways a static banner cannot: a slow load, a blank frame, an asset that did not fetch, or a webview that behaves differently from your test browser.
What to watch here is a consistency check rather than a rate. If impressions look healthy and nothing downstream moves, suspect delivery or loading before you blame the creative. A playable that never rendered looks exactly like one nobody wanted to touch.
2. First interaction (the first tap)
This is the first moment a viewer does something: a tap, a drag, a swipe. The ratio of first interactions to impressions tells you whether the opening earns attention and communicates what to do.
A low first-tap ratio usually comes from one of these:
- The first frame is unclear. The viewer cannot tell it is interactive. A static-looking screen with no cue reads as a video or a banner.
- The prompt is vague. "Tap to play" says less than a visible hand pointing at the exact object to move.
- The opening is slow. If nothing happens for a while, people scroll past or close it.
- The hook does not match the audience. A hardcore strategy opener shown to casual puzzle players will be ignored however clean the UI is.
A match-3 example: if the board is visible but the first move is not indicated, many viewers will treat it as a screenshot. Highlighting the single valid swap often changes the opening more than any visual polish would.
The design side of this is covered in Playable Ad Tutorial Design, which argues for teaching the mechanic with one action.
3. Engagement depth
Between the first tap and the end there is a middle: how many actions people take, and whether they keep going. Not every setup exposes this, but where it does, it is the most honest signal about the mechanic itself.
If people tap once and stop, the first action was confusing or unrewarding. If they keep going until the end, the loop works. Compare drop-off by action, not by elapsed time. A steep fall at a particular step means that step is the problem: an unclear instruction, a failure state that arrives too early, or a difficulty spike.
4. Completion
Completion means the viewer reached the designed end of the playable and saw the end card. Two different problems hide behind a weak completion ratio:
- People leave mid-play. The mechanic was not satisfying, the playable ran too long, or a step felt like a chore.
- Completion is not reachable. A bug, a dead end, or a win condition that is not obvious. This is a build problem, not a creative one. Always play the exported file from start to finish on a real device before blaming the audience.
The opposite failure exists too. If almost everyone completes and installs stay flat, the playable may be too easy or too short to build any desire for the real product. A fitness app that lets you "finish a workout" with a trivial gesture teaches nothing about why you would download it.
5. End-card click
The end card is where intent turns into action. Compare end-card views to clicks on the call to action. If people finish but do not click, the problem sits at the end:
- The button is easy to miss or competes with other elements.
- The payoff does not match the promise. They won a level and the end card shows something unrelated.
- The call-to-action wording is generic and does not say what happens next.
In a networked setup, the click also has to be wired to the right exit or store call. A broken or misrouted click looks exactly like a lack of interest, so test it in the network's own preview, not only locally. For Google Ads specifically, What Is Exit API in Google Ads? explains how that click is supposed to be handled.
6. Store page and install
After the click, the ad is no longer in control. The store listing takes over, and the remaining drop-off belongs to your icon, screenshots, rating, description and the user's patience.
This is the stage people most often misdiagnose. A playable can do its job perfectly and still lose the install because the store page contradicts it. If the ad promised a calm farming loop and the first screenshots show combat, the click will not convert. Check that the opening screenshots echo the playable's best moment.
Also allow for attribution delay. Installs are often reported later than the impressions that caused them, and different networks and measurement tools use different attribution windows. Do not judge a fresh creative on the first hours of data. Compare creatives over the same period and the same reporting setup.
Reading ratios, not raw counts
Raw counts depend on volume. Ratios between neighbouring stages tell you which decision is failing. Here is how a few common patterns read:
- Strong first tap, weak completion. The opening works, but the mechanic or length does not. Shorten the playable or make the middle more rewarding.
- Weak first tap, strong completion among those who start. The playable is fine but the opening hides it. Fix the first frame and the prompt.
- Strong completion, weak end-card click. The experience is good and the end card is not converting it. Rework the payoff and the button.
- Strong end-card click, weak installs. Look at the store page, the audience match, and whether the click lands where it should.
- Everything looks fine, installs are fine, quality is poor. The playable may be attracting the wrong people by promising something the product does not deliver. Check retention and in-app events, not just installs.
Change one stage at a time. If you rewrite the opening and the end card together, you will not know which change moved the result. The approach is laid out further in Playable Ad Creative Testing: A Practical Framework.
What you can actually measure
Funnels look tidy on a whiteboard and messy in dashboards. Networks differ in which events they expose. Some give you impressions, clicks and installs and nothing in between. Others let you fire custom events for first interaction and completion, or report engagement for you.
Before designing a test, answer these:
- Which events does this network report, and under what names?
- Is "click" a tap on the call to action, or any tap on the ad?
- Can you fire your own events from inside the playable, and are they allowed on this network?
- Do two networks define "completion" the same way? Usually they do not.
If a stage is not measurable, do not pretend. Use a proxy, and label it as one. For example, compare two versions of the opening using install outcomes alone and accept that you cannot see where the difference arose.
Be careful comparing funnels across networks, too. Placement, audience and the way an ad is shown all change behaviour. A ratio that looks weak on one network may be normal on another. Compare like with like, or compare a new creative against an existing one on the same network.
Diagnose, then decide
Early-funnel metrics are useful because they arrive quickly and point at specific parts of the creative. But they are proxies. A playable can raise first taps by being flashy and lose installs by overpromising. It can raise completion by being easy and fail to make anyone want the full product.
Use the funnel to decide what to change, and use installs and post-install behaviour to decide whether the change was worth keeping. When performance falls after a long run, check whether the cause is the creative wearing out or something in the funnel. Ad Creative Fatigue: Spot It Early and Know What to Swap covers the first case.
A short checklist before you read any numbers
- Play the exported playable on a real device, start to finish, in the network's preview.
- Confirm the call to action reaches the right store page.
- Confirm which events the network reports and how it defines them.
- Keep the audience, placement and period the same when comparing two versions.
- Wait for attribution to settle before drawing conclusions.
Most funnel "problems" that turn up in the first days of a test are one of these basics rather than a creative issue.
Where to go from here
Reading a funnel gets easier once you have seen several playables side by side. You can try live examples at /demos to see how openings, tutorials and end cards differ. If you want to build your own, PlayableRun turns a store listing into a playable and exports a Google Ads-ready ZIP, which you can test on a device before spending anything. A free account is at /signup.
Frequently asked questions
›What are the main playable ad funnel metrics?
The core stages are impression, first interaction, engagement depth, completion, end-card click, store page visit and install. Not every network reports all of them, so check which events your network exposes before planning a test.
›What does a low first-tap rate usually mean?
Usually the opening fails to signal that the ad is interactive or does not show what to do. Common causes are a static-looking first frame, a vague prompt, a slow start, or a hook that does not suit the audience.
›Is a high completion rate always good?
No. If nearly everyone finishes but installs stay flat, the playable may be too easy or too short to build desire for the real product. Completion shows the experience worked as a piece of content, not that it persuaded anyone.
›Why can a playable lose installs after a good end-card click?
After the click, the store listing takes over. If the icon, screenshots or description contradict what the playable promised, people leave. A broken or misrouted click can also look like a lack of interest.
›Should I optimise early-funnel metrics or installs?
Treat early-funnel metrics as diagnostics and installs, plus what installed users do afterwards, as the judge. Improving a middle stage is only worth it if it carries through to the end.
Turn your store listing into a playable
Paste an App Store or Google Play link and get an HTML5 playable you can test in minutes. The first one is free.
Keep reading
- User acquisition
Ad Creative Fatigue: Spot It Early and Know What to Swap
7 min read
- User acquisition
IPM vs CTR Playable Ads: Which Metric Actually Moves
7 min read
- User acquisition
Indie Game User Acquisition Without a Creative Team
7 min read