Google App Campaigns: How to Optimize for Users Who Actually Stay
Install volume is the wrong primary metric for app campaigns.
It's the number that looks good in a reporting dashboard and means the least for actual business outcomes. A campaign optimized purely for cheap installs will consistently bring in users who open the app once and disappear. When you factor in what you paid to acquire someone who generated zero sessions after day one, cheap installs are expensive.
The metric that matters is D7 retention — the percentage of users installed on day zero who are still active on day seven. According to Adjust's Mobile App Trends report, average D1 retention across app categories is around 25%, and D7 drops to roughly 10%. Campaigns that beat those benchmarks are working. Campaigns that track below them are buying the wrong users, regardless of what the CPI looks like.
Here's how to set up campaigns for quality, not just volume.
Before campaign setup: connect the right data
This is the step most teams rush past and regret later.
Connect your Google Play Console or App Store Connect to Google Ads. Implement Firebase for event tracking — not just installs, but the in-app events that indicate genuine engagement: completed onboarding, first core action, day-7 active session, subscription start, purchase.
Without in-app event data, Google's algorithm has one optimization signal: installs. Give it better signals and it finds better users. This is the single highest-leverage change in any app campaign setup. Everything else is marginal by comparison.
Campaign types and when to use each
| Campaign type | Optimizes for | Best used when |
|---|---|---|
| App Install (UAC) | Install volume at target CPI | Building initial user base, testing creative |
| App Engagement | Re-engagement of existing users | You have install volume but need to recover dormant users |
| Video Reach | Impressions, unique viewers | Launching into a new market, brand awareness phase |
| App Action (tROAS) | In-app conversions at target ROAS | You have 50+ conversions on a specific in-app event |
Start with App Install campaigns. Switch to optimizing for in-app actions once you have enough conversion data — Google requires roughly 50 conversions on a specific event before its algorithm can optimize reliably toward it.
Bidding: what to set and why
Target CPI to start. Set it at what you can afford based on the lifetime value of an average retained user — not just any install. If a D30 retained user is worth $12 to your business and you historically retain 15% of installs to D30, your install is worth $1.80. Start your target CPI near that number.
The most common mistake: setting CPI too low to win meaningful impressions, then concluding that app campaigns "don't work." The campaign is just starved for budget and reach.
Switch to Target CPA for in-app actions once you have the conversion volume. This is where campaign quality typically improves significantly — you're now telling Google to find people likely to do the thing that matters, not just people likely to install.
Don't touch bidding frequently. Google's learning period after a significant bid change is 2–4 weeks. Adjusting every few days resets the learning window continuously and prevents the algorithm from stabilizing.
Creative that Google's algorithm can actually use
Universal App Campaigns test combinations automatically. Your job is to give the system enough variation to find something that works.
Minimum inputs worth providing:
| Asset type | Quantity | What makes it effective |
|---|---|---|
| Headlines | 5 | Lead with user outcome, not feature name |
| Descriptions | 5 | Each addresses a different objection or use case |
| Images | 3+ | Landscape, square, and portrait formats all provided |
| Videos | 1–3 | Value proposition in first 5 seconds, before skip option |
The first five seconds of any video carry disproportionate weight. State clearly who this is for and what they'll get — before the skip button appears. "If you manage a remote team and your async communication is chaos, this is for you" is a first five seconds. A logo animation followed by a product tour is not.
Videos with a clear direct response hook (Google's own research shows the first five seconds determine whether viewers stay) consistently outperform brand-awareness style creative in install campaigns.
Targeting worth using
Custom intent audiences — Build audiences around specific search queries people have used on Google. Someone who searched "best project management app for remote teams" last week is a meaningfully different audience than a broad "productivity" interest category. This is the most underused targeting lever in most app campaigns.
Placement targeting — Show ads on specific YouTube channels or apps. Requires research but produces the highest audience relevance.
Lookalikes from your customer list — Upload your best user emails and target similar audiences. Available under Audience Manager.
What to exclude actively: Your existing subscribers (if the goal is acquisition). Placements where your ad will appear next to irrelevant content. Low-quality Display Network placements — use placement exclusion lists.
The audit that tells you if a campaign is working
Pull three numbers and put them side by side:
- CPI from Google Ads dashboard
- D7 retention rate from Firebase or your analytics tool
- In-app conversion rate (what percentage of installs reach your key event)
If CPI is low but D7 retention is below 8%, the campaign is bringing in the wrong users — most likely through broad targeting or creative that overpromises. The fix is usually targeting refinement or creative that more accurately represents the actual product experience.
If CPI is high and D7 retention is good, the campaign is working — the unit economics just need to close. Either the LTV needs to improve or the CPI needs to come down through better creative.
The combination to shoot for: CPI at or below your LTV-justified ceiling, D7 retention at or above category average. When you have that, scale the budget.
We manage app campaigns for clients as a certified Google Partner — and the conversation that actually moves results is almost never about the campaign settings.
It's about what happens after the install, and whether the app experience justifies what we're paying to get people there.
If your installs aren't turning into retained users, that's the conversation →
