Short answer: Google App campaigns show you results at campaign, ad group and asset level, plus a few supporting reports: top asset combinations, placements (impressions only), network segments and, on iOS, SKAdNetwork. The asset report rates each asset Low, Good or Best against other assets of the same type in the same campaign, using percentile rankings based on average daily cost and the last 14 days of data. A rating tells you how an asset compares inside Google, not whether the people it brings in go on to pay, so make in-campaign decisions from Google’s reports and business decisions from your own revenue data.
Page Contents
What reports do Google App campaigns give you?
App campaigns build ads automatically from the headlines, descriptions, images, videos and HTML5 assets you upload, and serve them across Google Search, Google Play, YouTube, the Google Display Network and Discover, according to Google’s overview of App campaigns. The reports are built around assets. The ones worth knowing:
- Asset report (per ad group): clicks, impressions, cost and conversions for the ads that contain each asset, with a segment option to see how networks and conversion types contributed. Rows overlap because an ad can use several assets; Google explicitly notes that some impressions count in more than one row (Google: asset reporting).
- Cross-campaign asset report: each active asset with impressions in the last 30 days, how many ad groups use it, and the share of its impressions spent at each rating (same page).
- Top combinations report: the 6 best-performing combinations in each of three groups (with an image, with a video, text only), with an ad preview. It can be empty for new or low-volume campaigns (Google: top combinations).
- Placement report: built in Report editor. Impressions are the only metric, and low-activity placements are grouped into “Total: other” (Google: placement reports).
- SKAdNetwork report (iOS): campaign-level installs and conversion values from Apple’s postbacks, under Reports > Goals and conversions (Google: SKAdNetwork reporting).
Do not sum asset rows as if they were independent ads: an ad can contain several assets, and the report also leaves out auto-generated assets. Those are reasons asset totals need not match campaign totals (Google).
How do Low, Good and Best asset ratings work?
Google’s asset reporting page defines the statuses:
| Rating | What Google says it means |
|---|---|
| Pending | Still being processed; no rating yet. It does not affect serving. |
| Learning | The campaign needs more data before it can rank the asset. |
| Low | One of the lowest performing relative to other assets of its type. |
| Good | Performing well relative to other assets of its type. |
| Best | One of the best performing relative to other assets of its type. |
| Unrated | The system cannot rank it because there are too few assets. |
Four details matter:
-
They are based on average daily cost. Google describes them as percentile rankings by this measure, not rankings of standalone conversion rate, return or causal contribution.
-
They are relative. In the ad group report, the comparison is with assets of the same type in the same campaign. If every video you uploaded is weak, one of them can still be rated Best.
- They use the last 14 days and are always current. Changing the date range does not change the rating column, and ratings move with user behaviour and network mix.
- Google publishes no thresholds. There is no official impression or conversion count at which an asset leaves Learning.
What should you do with Low, Good and Best assets?
Google’s own recommendations are short: replacing a Low asset could help, keep Good assets and add more, and add more assets like your Best ones (asset reporting). Its creative guidance adds that once you are at the asset limit, replace Low assets gradually rather than all at once. The limits are up to 5 headlines and 5 descriptions, and up to 20 total of each asset type: images, videos and HTML5 per ad group. HTML5 is for games (Google: asset types).
Two things the ratings won’t tell you:
- Whether the asset independently brings valuable users. Best is not a revenue-quality verdict. Check campaign or ad-group outcomes against your revenue data at the granularity your measurement supports; do not assume an MMP can identify the people caused to convert by one asset.
- Whether you need a new idea or a new version. Swapping a Low image for a slightly different image is a variant. If every asset in a campaign is converging on the same message, the next gain usually comes from a new creative direction, tested with several ads across angles, hooks and formats. Creative is the main lever in App campaigns because Google does the targeting; what creative assets you need for Google App campaigns covers what to make, and why targeting is the wrong question covers why.
How do you see which combinations and placements are working?
The top combinations report (Campaigns > your App campaign > Assets > Combinations) shows which images, videos and text Google pairs most successfully. Only the top 6 per group appear, so a missing combination just means others did better in that period (Google).
For where the ads ran, segment your campaign by network for spend and conversions, and use the placement report for the apps, sites and videos that served impressions. The placement report is a delivery-location report, not a placement-level cost-per-conversion ranking.
When can you trust what the reports say?
Google’s guidance on evaluating App campaign performance calls at least 100 conversions ideal before evaluating cost per conversion and explains that daily costs can vary around a monthly average. That is Google’s recommendation, not a guarantee that 100 conversions make a comparison conclusive. It also warns about conversion delay: Google credits a conversion to the day of the click, so recent click-date data can be incomplete and cost per conversion can look higher before delayed conversions arrive. Check what share of conversions arrive within 7, 14 and 30 days of a click, leave the most recent days out of your read, and make sure your conversion window covers the delay.
Volume matters too: a campaign gives a usable read faster with more conversions a day, ideally 30, and 10 is good. Google’s own best-practice budget is at least 50x your target cost per install, or 10x your target cost per action, per day. You may also have heard you need 50 conversions a week before anything works; that is really old guidance we would avoid.
Why don’t Google’s numbers match your MMP?
They count different things. Google has its own attribution and reports view-through conversions separately, and its optimisation tips suggest adding view-through conversions when you calculate an adjusted cost per install. Your MMP applies its own attribution rules across every channel, so a user Google credits may be credited elsewhere.
iOS widens the gap. Google’s iOS measurement includes SKAdNetwork, on-device measurement and integrated conversion measurement where implemented. It also shows modelled conversions where it can’t link a conversion directly to an ad; it does not currently support AdAttributionKit for attribution (Google: iOS best practices). Its SKAdNetwork report also redistributes missing conversion values with a model, and Google says the results can differ from third-party SKAN reports (Google). For more on that side, see how to fix mobile attribution on iOS.
The practical split: use Google’s reports for decisions inside Google (which assets to replace, whether a target is holding), and use your MMP or revenue data to decide how much budget Google deserves next to your other channels.
Frequently asked questions
What does “Learning” mean in the App campaign asset report? It means the campaign needs more data before it can rank that asset against others of the same type. Google does not publish how much data that takes.
What is the difference between Pending and Unrated? Pending means Google Ads is still processing the asset. Unrated, in the cross-campaign report, means there are too few assets of that type for Google to rank it.
Why doesn’t my asset spend add up to campaign spend? One ad can contain several assets, so rows overlap rather than describing independent ads. Auto-generated assets are also excluded. Use campaign totals for overall spend.
Can you see performance by placement in Google App campaigns? Only impressions. The placement report shows impressions by app, site or video, and groups low-activity placements into “Total: other”; use network segments for cost and conversions.
How long should you wait before judging a Google App campaign? Google calls at least 100 conversions ideal before evaluating cost per conversion. It is not a statistical guarantee. Read a period long enough to account for your conversion delay and leave out recent, incomplete click cohorts.
