Should Google App Campaigns Be Your Game's Main UA Channel? 2026 Campaign Types, Bidding Thresholds, Budget Floors, What You Can't Control, and a Pilot Plan
Google App Campaigns take a few lines of text, images and videos, a bid, a budget, and your languages and locations. Google then decides which ad combinations to show and where, across Google Search, Google Play, YouTube, the Display Network, and Discover. For Android game studios it is close to the default channel. In the 2025 edition of the AppsFlyer Performance Index, published in December 2025, Google Ads again ranked first for Android gaming.
But "everyone runs it, so we switch it on" is a different decision from "we put our main budget behind it." App campaigns are heavily automated. In practice, the advertiser chooses five things: the campaign type, the optimization event, the bidding goal, the budget, and the creative assets. Choose those five badly and Google's machine learning will very efficiently hit the wrong goal. Based on Google Ads Help documentation as of October 2026, this article sets out the requirements and numbers to check before you adopt App campaigns or scale them, and how to make the call with a short pilot.
We compared install optimization and event optimization across channels in Buying What Happens After the Install: How to Structure App Install Campaigns for a Mobile Game in 2026. This article goes deep on Google alone.
What exactly are you buying with Google App Campaigns?
With App campaigns, you are not buying specific placements. You are buying Google's automated optimization toward the goal you choose (installs, in-app actions, or value) within the price you set.
According to Google Ads Help, the advertiser supplies some text, a starting bid and budget, and the languages and locations, and Google Ads tests different ad combinations and serves the best-performing ones. Not having to build separate ads for each network and format is presented as a benefit. Read the other way, it also means the advertiser does not decide how much goes to each network. Ads can appear on Google Search and search partners, Google Play (search results, related apps sections, and the homepage), YouTube, the Display Network (Gmail, other apps, and mobile websites), and Discover.
So the buying decision is less "is this a good channel?" and more "can our game give Google's optimization the right signal, in sufficient quantity?" The choice of optimization event determines the quality of that signal. Budget and conversion volume determine its quantity. The sections below take these two in turn.
Install, engagement, pre-registration: when do you use each campaign type?
Google App Campaigns come in three types: installs (ACi) for new users, engagement (ACe) for re-engaging existing users, and pre-registration before launch. Each has different entry requirements.
Type | Purpose | Entry requirements (per Google Ads Help) | Bidding goals |
|---|---|---|---|
App campaigns for installs (ACi) | New installs | No strict requirements. Setting the right price and suitable asset types is recommended | tCPI, tCPA, tROAS, Maximize conversions, Maximize conversion value |
App campaigns for engagement (ACe) | Re-engage users who installed or used the app (high-value lapsed users, in-app purchases, users who installed but never opened, live events and sales) | At least 50,000 installs, deep links and audience lists, conversion tracking via Firebase or an app attribution partner (AAP), and a privacy policy that follows Google's EU user consent policy | tCPA, tROAS |
App campaigns for pre-registration (Android only) | Pre-registrations before a Google Play launch | APK uploaded to a Play Console release track, launch within 90 days of opening pre-registration, the game must not already be installable or purchasable in the targeted country, child-directed apps excluded | Target cost per pre-registration (tCPpre) only |
Be careful with the engagement campaign install requirement, because Google's documents disagree. The December 2020 global launch notice required 250,000 installs. The current "different types of App campaigns" page says 50,000. If you run a small or mid-sized live game, check against the 50,000 figure first, then confirm in your own account that you can actually create an engagement campaign. Engagement campaigns only serve with an attached audience segment such as Customer Match or a Firebase list, so the data work to build lapsed-user lists comes first.
Pre-registration campaigns matter especially for game studios. Google recommends turning on pre-registration at least two weeks before pre-launch marketing begins. It says three to six weeks is typical, and nine to twelve weeks when you expect high interest. Ads show only on Android phones and tablets, on Google Play, YouTube (in-stream), and other apps and mobile websites. You can set a pre-registration reward only once for the lifetime of the campaign, and you cannot edit or delete it after creating it. When the game launches in a targeted country, ads stop there and continue in the remaining countries. We covered the difference between iOS pre-orders and Google Play pre-registration, and how to design the rewards, in App Store Pre-Order vs Google Play Pre-Registration.
Which bidding strategy, and when? The thresholds for tCPI, tCPA, and tROAS
Choose the bidding strategy by how much conversion signal your game can send Google, and move up one step only as volume builds.
Bidding strategy | Optimizes for | When to use | Conditions Google states |
|---|---|---|---|
tCPI (target cost per install) | Installs | Early after launch, when in-app event volume is low | None. For the "users likely to perform an in-app action" option, bid 20% above your baseline tCPI |
tCPA (target cost per action) | One selected in-app event | Once events such as tutorial complete, level reached, or first purchase are accumulating | Select only one event (multiple events create a blended target). Set tCPA 20% above the CPA you actually observed |
tROAS (target return on ad spend) | Conversion value (revenue) | Once purchase or ad revenue values arrive steadily | Typically 30+ conversions in 30 days. When moving from tCPA, three to four weeks of conversion value history. Dynamic revenue values passed via Firebase or Google Play codeless conversions |
tCPpre | Pre-registrations | Before launch | Pre-registration campaigns only |
The three work like a ladder. Start with tCPI to establish a baseline install cost. Move to tCPA as in-app events accumulate, and to tROAS once revenue values are stable. Google strongly recommends running a tCPA campaign before tROAS to learn what ROAS is achievable. It also says to set the initial tROAS from the ROAS actually achieved by a tCPA campaign with the same app ID, the same event, and the same location and language targeting. Google gives an example: if you bid a $6 tCPA and your actual CPA was $3, apply a 50% modifier to the ROAS you calculated.
The common mistake when choosing a tCPA event for a game is picking too deep an event, such as first purchase, too early. If only a few conversions come in each day, the machine learning has too little to learn from. An event that is too shallow, such as session start, produces plenty of conversions but behaves much like install optimization. Google itself notes that frequent actions such as session start generate more conversions than usual. In practice, the key is finding a middle event: one that correlates strongly with purchasing or long-term retention and still fires at least dozens of times a day. Google recommends iOS bids of 1.5 times Android, so keep that gap in mind when you compare the two operating systems within one budget.
How much budget do you need before results mean anything?
Google sets a minimum daily budget multiple for each campaign type, and a budget below that multiple is likely to produce a campaign that never finishes learning.
Google Ads Help's best practices for App campaigns sets these daily budget floors as multiples of the bid: at least 50x for install campaigns on tCPI and for pre-registration campaigns, at least 10x for install campaigns on tCPA, and at least 15x for engagement campaigns on tCPA. The table below applies those multiples for illustration. The bids are assumptions.
Campaign | Assumed bid | Minimum daily budget | 30-day equivalent | Expected conversions per day |
|---|---|---|---|---|
Install campaign, tCPI | $1.50 | $75 (50x) | $2,250 | about 50 |
Install campaign, tCPA (tutorial complete) | $6.00 | $60 (10x) | $1,800 | about 10 |
Engagement campaign, tCPA (lapsed user returns and reaches a level) | $4.00 | $60 (15x) | $1,800 | about 15 |
Pre-registration campaign, tCPpre | $0.80 | $40 (50x) | $1,200 | about 50 |
Read the learning rules alongside the budget. Google warns that changing an in-flight campaign before its first 100 conversions register may disrupt learning and hurt performance. At about 10 conversions a day, as in the tCPA example above, reaching 100 takes about ten days, so budget on the assumption that you will leave the campaign alone for at least two weeks. After that, Google advises against drastic changes of more than 20% to budget or CPI. The tROAS guidance says to expect a 7 to 14 day learning phase after major budget or target changes. Evaluate over a 14 to 30 day window, leaving out the most recent days while conversions are still coming in. Google also notes that Maximize conversions campaigns with small budgets may not get any conversions.
In short, even a "test" of Google App Campaigns needs roughly $1,500 to $2,500 a month per campaign, plus two to four weeks of patience. If that is hard to fund at your stage, concentrate the budget in a single campaign rather than spreading it across several.
What you can control, and what you can't
In App campaigns, the levers in the advertiser's hands are creative assets, the optimization event, the bidding goal, the budget, and audience signals. Network allocation and impression-level decisions are left to Google.
Area | What you can do | What you can't do, or can only do partly |
|---|---|---|
Creative | Provide four lines of text, 20 images, and 20 videos in a variety of aspect ratios (Google's recommendation). At least one approved text, image, and video asset per ad group | Specify which combination runs on which placement |
Targeting | Languages, locations, audience signals (app user and Firebase lists recommended) | Signals are hints, not restrictions on who sees the ads. Install campaigns don't apply demographic exclusions |
Placements | Account-level placement exclusions (Display, YouTube, search partners) | Budget or bid allocation by network |
Optimization | One event, target cost or ROAS, conversion window | Fine-tuning during the learning period |
In Google's words, audience signals give Google AI additional information about the type of users most likely to install your app, and they are optional. Adding a list of existing payers can help with the cold start in early learning. It does not work as an instruction to "show this only to these gamers."
Reporting has limits too. The asset report rates each asset Low, Good, or Best against other assets of the same type. That is a relative rating, and Google says the labels can change. A Low image and a Low headline cannot be compared with each other. Only assets with impressions in the last 30 days appear in the report. Conversion counts that don't match your MMP or Play Console are normal. Google Ads counts only installs that came through Google Ads. In pre-registration campaigns, a pre-registration still counts as a conversion if the user later unregisters, and a user who unregisters and pre-registers again through Google Ads can be counted twice. App status changes can lag by 12 to 24 hours, so daily counts can also drift. We explained why every dashboard shows a different install count in Why Every Dashboard Shows a Different Install Count.
Because of these limits, judge Google App Campaigns on MMP cohort data, such as D7 retention and D7/D30 ROAS for users installed in the same period, rather than on the Google Ads screen. Whether Google is also taking credit for users who would have arrived organically can only be checked with a holdout or a geo-split experiment. For the experiment design, see How to Design an Incrementality Test for Mobile Game UA.
When Google App Campaigns should be your main channel, and when they need a complement
Google App Campaigns become the main channel for games with a high Android share, enough volume to send at least dozens of meaningful in-app events a day, and a steady supply of creative.
Situation | Role of Google App Campaigns | Why |
|---|---|---|
Android-centric, daily budget in the hundreds of dollars or more, enough event volume | Main channel (volume anchor) | Meets the budget multiples and learning conditions, and can move up to tCPA and tROAS |
Before launch | Run alongside as a pre-registration campaign | Only tCPpre is available, and behavior after pre-registration is not an optimization target |
Just after launch, low in-app event volume | Use tCPI for a baseline, and supply event depth through other channels | Deep events such as first purchase or long-term retention take time to generate enough learning data |
Reactivating lapsed users | Engagement campaign, if you have 50,000+ installs | Deep links, audience lists, and conversion tracking must be in place first |
Mid-core and hardcore games with distinct genre audiences | Main channel, but needs help on targeting precision | Audience signals are only hints, so you cannot specify genre preference directly |
Daily budget in the tens of dollars | Concentrate on one campaign, or hold off | If you can't meet the 50x or 10x floor, learning never completes |
Google's weak spots come down to two areas. The first is periods with little learning data, such as pre-registration and the first weeks after launch. The second is audiences Google's signals cannot specify directly, such as "gamers who actually play this genre for a long time." Running a complementary channel that covers these two gaps also gets better conversion signals to the Google campaign sooner. We set out the criteria and RFP questions for comparing channels in How to Choose a Mobile Advertising Platform for Your Game in 2026.
How do you design a four- to six-week pilot?
The point of the pilot is to find out whether Google can learn your target event within your target cost. To do that, fix the event, the budget, and the evaluation criteria before you start.
Phase | Timing | What to do | Pass criteria (example) |
|---|---|---|---|
Preparation | Before launch | Connect Firebase or MMP conversions. Choose one optimization event that fires at least dozens of times a day and correlates strongly with D7 retention. Prepare four lines of text, plus images and videos in multiple aspect ratios. Document the evaluation metrics and target values | Conversions arrive in both Google Ads and the MMP |
Learning | Weeks 1–2 | One campaign, budget at or above the multiple for its type, no changes until the first 100 conversions | Daily conversions at 70%+ of target volume |
Stabilization | Weeks 3–4 | Replace Low-rated assets from the asset report, adjust budget and target within 20% | Holds within ±20% of target CPA |
Decision | Weeks 5–6 | Compare D7 retention and D7 ROAS on a 14 to 30 day cohort, excluding the most recent days; add a geo holdout if possible | Scale if equal to or better than cohorts from other channels |
The most important comparison when reading the result is against cohorts from other channels over the same period, on the same event. If Google's CPA is low but its D7 retention is lower than other channels', it has probably learned a shallow event. Change the event to a deeper one and let it learn again.
Playio is one of the channels you can run as that comparison. Playio is an Android-based community of 5 million gamers who love games. It uses genre preferences, play history, and in-game behavior data to match games with gamers whose tastes fit. The default pricing is CPI, and Playio also supports CPE tied to in-game actions such as tutorial completion, reaching a level, or hitting a playtime goal. You can set the same event as your Google tCPA campaign and compare the cohorts side by side. Before launch, Playio's pre-registration products (Step-Up, Premium, and Premium PLUS tiers) can gather pre-registrations alongside a Google pre-registration campaign. They are provided free of charge depending on campaign spend. Separately securing players who complete deep events during the first few weeks, while the Google campaign is still building learning data, also gives you a basis for setting audience signals and tCPA targets. We recommend agreeing on the comparison event and period at the pilot design stage.
You can find more details here. (https://playioadsen.oopy.io/bizdeck)
Key Takeaways
With Google App Campaigns, you are buying Google's automated optimization toward the goal you choose, not placements. What you choose is the campaign type, the optimization event, the bidding goal, the budget, and the creative. As of October 2026, Google Ads Help lists three campaign types: installs (no strict requirements), engagement (50,000 installs, deep links, audience lists, conversion tracking), and pre-registration (Android only, launch within 90 days, tCPpre only). Recommended daily budgets are at least 50x the bid for tCPI and pre-registration, 10x for install campaigns on tCPA, and 15x for engagement campaigns on tCPA. Leave a campaign alone until its first 100 conversions, then adjust within 20%. Move from tCPA to tROAS only after about 30 conversions in 30 days and three to four weeks of conversion value history. You can't control network allocation, audience signals are hints, and Google Ads counts conversions differently from your MMP, so make decisions on MMP cohorts and incrementality tests. For games with a high Android share and enough event volume, App campaigns can be the main channel. For pre-registration, the first weeks after launch, and gamers with distinct genre tastes, validating them in a pilot alongside a complementary channel is the safer path.
For inquiries about Playio's advertising solutions, reach out at: [email protected]