Google App Campaigns, explained: bidding, assets, learning and when to get help
App Campaigns leave you only a few levers: bid strategy, event, budget and assets. Here's how to set each one, and when an agency is worth it.
- Google App Campaigns give you few levers, so bid strategy, event choice, budget and assets decide results.
- Google recommends a daily budget of at least 50x target CPI or 10x target CPA for in-app actions.
- Optimize toward one event per campaign, the deepest one with enough volume.
- Avoid changes before the first 100 conversions and avoid cycling bids up and down.
- An agency pays off when spend is meaningful, events need fixing or you lack creative capacity.
Google App Campaigns are simple to launch and easy to get wrong. You give Google a budget, a bid target, an event and some assets, and its systems decide where your ads run across Search, Google Play, YouTube, Discover and display inventory. Because so few levers are left in your hands, each one matters a lot. That's the real case for a Google App Campaigns agency: not access to secret settings, but discipline about the few settings that exist.
This guide explains how App Campaigns work, how to choose between target CPI, target CPA and target ROAS, how to handle assets and the learning period, and when it's worth bringing in outside help.
How Google App Campaigns work
Unlike Search campaigns, App Campaigns have no keywords, placements or manual audience targeting in the usual sense. You provide:
- A goal and bid strategy: installs, in-app actions, or in-app action value.
- A budget and target: for example a target cost per install or per in-app action.
- Locations and languages.
- Assets: headlines, descriptions, images, YouTube videos and optional HTML5, which Google combines into ads for each placement.
- Conversion data: in-app events from Firebase or a mobile measurement partner, which the system uses to find similar users.
Google then runs auctions across its inventory and learns which combinations of user, placement and asset produce your chosen event at your target cost. Your job, or your agency's, is to feed it good signals and avoid disrupting the learning.
Choosing a bid strategy: tCPI, tCPA or tROAS
The bid strategy decides what Google optimizes for. Picking the wrong one is the most common and expensive mistake we see.
| Strategy | Optimizes for | Use it when | Main risk |
|---|---|---|---|
| Target CPI (installs) | Volume of installs at your target cost | Brand-new app with no event history, or you need install volume to train other campaigns | Cheap installs from people who never open the app again |
| Target CPA (in-app actions) | A chosen event, such as signup, trial or purchase | You have steady event volume and know what an action is worth | Too little event volume makes delivery unstable |
| Target ROAS (action value) | Revenue relative to spend | Purchases with varying values and solid revenue tracking | Needs value-passing events and enough purchase data; restrictive targets throttle scale |
| Maximize conversions | Most installs, actions or value within budget | Testing or when you'd rather control budget than cost | Cost per result can swing |
According to Google's App campaign bidding documentation, target ROAS requires the Google Analytics for Firebase SDK and events that pass a dynamic revenue value. Google also recommends running a target CPA campaign before moving to target ROAS, so you have a realistic starting point.
Budget relative to your target
Underfunded campaigns can't learn. Google's App campaign best practices recommend a daily budget of at least 50 times your target CPI for install campaigns, at least 10 times your target CPA for in-app action campaigns, and at least 15 times target CPA for engagement campaigns. So a $40 target CPA implies a daily budget of at least $400. If your budget can't support that for a deep event, optimize for an earlier, more frequent event first.
Picking the right in-app event
Event choice is a trade-off between volume and value. Optimizing for a purchase finds better users, but if you only get a handful of purchases a day, the system has too little to learn from. A practical approach:
- List the funnel: install, signup, key action (for example first lesson, first search, KYC complete), trial or first purchase.
- Pick the deepest event that reliably happens often enough to meet the budget rule above.
- Optimize toward one event per campaign. Google warns that targeting several actions creates a "blended" target that's harder to control.
- Move one step deeper once volume allows, as a new campaign rather than an edit to the old one.
Conversion windows matter too. Google advises windows under 30 days, long enough to capture roughly 90% of your conversions. If most users purchase on day one, a short window gives the system faster feedback.
Asset strategy that actually moves results
Assets are the main creative lever. Per Google's asset specifications, you can add up to 5 headlines (30 characters each) and 5 descriptions (90 characters each), images in 1:1, 1.91:1 and 4:5 ratios, YouTube videos in landscape, portrait and square, and HTML5 ZIP files.
- Fill every format. Missing a video orientation means Google auto-generates one or skips placements, often YouTube Shorts and other vertical inventory.
- Write headlines that each stand alone. They're shown in combination, so each should make sense by itself: one benefit, one proof point, one offer.
- Read the asset report. Replace assets rated "Low" after they've had meaningful impressions, and keep Ad Strength at "Good" or better.
- Refresh in batches. Swap a few assets at a time on a regular schedule, rather than replacing everything at once and resetting learning.
Share your app and goals. We'll send a channel plan, starting budget and the first creatives we'd test — free, within 48 hours.
Get my free growth plan WhatsApp usRespecting the learning phase
New App Campaigns, and campaigns after big changes, go through a learning period while the system gathers data. Google's guidance is that changes made before the first 100 conversions can disrupt learning and hurt performance. In practice:
- Make bid and budget changes in small steps, and avoid cycling targets up and down.
- Judge results on several days of data, not on one day's cost.
- Launch new events, geographies or major creative pivots as separate campaigns where possible.
- Keep a change log so you can connect performance shifts to what changed.
Measurement setup: Android vs iOS
On Android, Firebase or an MMP gives Google good event data. On iOS, Apple's privacy rules mean less user-level signal: attribution relies on SKAdNetwork and AdAttributionKit for many users, and results arrive delayed and aggregated. Budget and targets for iOS usually need more patience and a separate read on performance. Our SKAdNetwork iOS attribution guide explains the moving parts.
Either way, check that the events Google sees match what your MMP and backend report. A double-counted or missing purchase event will quietly steer your bids in the wrong direction.
Common mistakes in App Campaign accounts
- Too many small campaigns. Splitting a modest budget across many countries or events leaves each one starved of data. Consolidate until each campaign can meet the budget rule.
- Optimizing for an event nobody uses later. "Registration" can look healthy while paying users stay flat. Check how the optimized event correlates with revenue before scaling.
- Judging Google by last-click numbers alone. App Campaigns reach people across YouTube and display who may install later. Compare MMP data, Google's reporting and your backend before cutting spend.
- Letting assets go stale. When the same handful of videos run for months, costs tend to creep up. Put refreshes on the calendar.
- Ignoring the store listing. Every App Campaign ad ends on your Play or App Store page. A weak listing raises your cost per install no matter how good the bidding is.
When to hire a Google App Campaigns agency
| Situation | In-house is fine | An agency usually helps |
|---|---|---|
| Spend level | Small tests while you validate the product | Meaningful monthly spend where a few percent of efficiency pays for the fee |
| Event and measurement setup | You have an engineer who owns Firebase or the MMP | Events are messy, missing or don't match the backend |
| Creative volume | You can produce fresh video and images every few weeks | No in-house creative capacity for App Campaign formats |
| Bid strategy progression | One campaign, one event, stable results | Moving from installs to tCPA to tROAS, across several countries |
| Regulated categories | Not applicable | Finance, health or other categories with certification or policy requirements |
A good agency should work inside your Google Ads account, explain every structural change, and report on cost per key in-app event, not just cost per install. If you run several channels, also read about choosing a mobile user acquisition agency, and our guide to app install campaign optimization for ongoing tuning.
Running App Campaigns with Appryze
Our founder has 12+ years running user acquisition on Google App Campaigns, Meta and Apple Ads for consumer and fintech apps, including leading growth for an Indian NBFC lending app at scale. We start with your event and conversion setup, then build the campaign progression from there, inside accounts you own, with weekly written updates. Learn more on our Growth page.
Frequently asked questions
What is the difference between tCPI and tCPA in Google App Campaigns?
Target CPI optimizes for installs at an average cost you set, so Google looks for people likely to install. Target CPA optimizes for a specific in-app action, such as a signup or purchase, at your target cost. tCPA usually finds higher-quality users but needs steady event volume and a larger budget relative to the target.
How much budget do Google App Campaigns need?
Google's best practices recommend a daily budget of at least 50 times your target cost per install for install campaigns, at least 10 times your target CPA for in-app action campaigns, and at least 15 times target CPA for engagement campaigns. Budgets below that make it hard for the campaign to learn.
How long is the learning phase for Google App Campaigns?
There's no fixed number of days. Google's guidance is that changes made before a campaign has recorded its first 100 conversions can disrupt learning and lead to poorer results. Avoid big bid, budget or event changes during that period, and judge results over several days rather than one.
When should I use target ROAS for app campaigns?
Use target ROAS when purchase values vary and you track revenue reliably. Google's documentation says it requires the Google Analytics for Firebase SDK and events that pass a dynamic revenue value, and recommends running a target CPA campaign first to find a realistic starting target.
Do I need an agency for Google App Campaigns?
Not always. If spend is small and you have someone who owns event tracking and creative, you can run them in-house. An agency tends to pay off when spend is meaningful, events are messy, you lack creative capacity, or you're progressing from installs to in-app actions and value across several countries.
Share your app and goals. We'll send a channel plan, starting budget and the first creatives we'd test — free, within 48 hours.
Get my free growth plan WhatsApp usKeep reading
More in Google App Campaigns Agency · All app marketing guides