Grow Your App With High-Quality Users at Scale
Installs are easy to buy. Users who register, come back next week and pay are not. As an app marketing agency, we run app user acquisition on Meta, Google, TikTok and Apple Search Ads, and we judge every campaign on Day-30 retention and in-app revenue, not on the install count.
App Marketing Performance
App User Acquisition That Focuses on Quality, Not Just Volume
Every ad platform will happily sell you cheap installs. Most of those people open the app once and never again.
We run Meta App Install campaigns, Google App Campaigns (UAC), TikTok App Ads and Apple Search Ads, with attribution through Firebase, Adjust or AppsFlyer so every platform is judged by the same numbers. Creative is tested separately for each platform, because a video that works on TikTok usually dies on Google’s Display inventory.
The number we care about is cost per user still active at Day 30. A $2 install that stays is cheaper than an 80 cent install that leaves before lunch.
Same $8,000, two ways to spend it. Sample cohort report for a fictional app.
| Cohort | Spend | Installs | CPI | Day 1 | Day 7 | Day 30 | Cost per D30 user |
|---|---|---|---|---|---|---|---|
| Optimised for cheapest install | $8,000 | 10,000 | $0.80 | 22% | 7% | 3% | $26.67 |
| Optimised for registration | $7,980 | 3,800 | $2.10 | 46% | 26% | 14% | $15.00 |
The cheap cohort looks better on every dashboard that stops at installs. It loses on the only line that pays your bills.
Four Channels, Each Bought Differently
Same app, four different auctions. Here is what each one is good for and what we tell it to chase.
Meta App Install
Facebook, Instagram, Audience NetworkAdvantage+ app campaigns do the targeting. Our job is the event we feed them and the creative. We usually switch Audience Network off at launch, because it loves cheap installs that never register.
Google App Campaigns
Search, Play, YouTube, Discover, DisplayYou hand Google text, images and video and it builds the ads. Good assets in landscape, square and portrait matter more than any setting. We replace “Low” rated assets every two weeks.
TikTok App Ads
For You feed, Pangle off by defaultCheap reach and fast learning, but videos burn out in a week or two. We plan a steady feed of native-looking, creator-style clips, and move budget the day a hook stops working.
Apple Search Ads
App Store search results and tabsPeople searching the App Store already want an app like yours. We split brand, category, competitor and discovery keywords into their own campaigns, and judge each on registrations in the MMP, not on taps.
Why Quality App Acquisition Matters
In-App Event Optimisation
We tell each platform to find people who register, finish a profile or buy, not people who install. The algorithm finds exactly what you ask for, so ask for the right thing.
Mobile Measurement Partnership
Firebase, Adjust or AppsFlyer set up as the single source of truth, with postbacks to every network. Platforms all claim the same install. The MMP decides who gets credit.
App Creative Strategy
Show the moment the app pays off, in the first two seconds. Separate cuts for TikTok, Reels and Google, plus store screenshots that match what the ad promised.
Apple Search Ads
The highest-intent iOS traffic there is: people typing your category into the App Store. Tight keyword groups, negative keywords and bids set by what each term earns.
Retention-Correlated Targeting
We compare Day-7 retention by country, platform, placement and creative, then move spend toward the segments whose users stay. Some cheap geos never retain. We stop buying them.
LTV-Based Scaling
Budget goes up when a cohort’s revenue curve says it will pay back, not when install numbers look good. That keeps growth profitable as spend climbs.
Optimise for the Event That Predicts Revenue
Ask for purchases on day one and the algorithm starves for data. Ask for installs forever and it fills your app with tourists. So we climb.
Pick the deepest event you can feed
Each platform needs a minimum number of conversions to learn. Below that, delivery wobbles and costs jump. We start with the deepest event that clears the bar, usually registration, and move one rung down the funnel when volume allows.
If no event clears the bar on your budget, we say so, and suggest a cheaper event that still predicts retention, such as finishing onboarding.
| Platform | How it optimises | Volume to aim for |
|---|---|---|
| Meta | App event or value optimisation in Advantage+ app campaigns | ~50 a week per ad set |
| Target CPA on an in-app action, then target ROAS | ~10 a day | |
| TikTok | App event optimisation, then value-based bidding | ~50 a week per ad group |
| Apple Search Ads | Keyword bids per tap, with a CPA goal | Judged in the MMP |
How We Measure iOS When Most Users Say No to Tracking
Since Apple’s App Tracking Transparency prompt, only the users who tap Allow can be tracked one by one. Everyone else arrives late, in aggregate, through SKAdNetwork. Planning for that delay is half of iOS app marketing.
MMP postbacks
Users who allowed tracking on iOS, plus Android installs through the Play Install Referrer. Near real time and user level. This is where cohorts and LTV come from.
SKAN and AdAttributionKit
Every attributed iOS install, but delayed and grouped. We design the conversion value schema so the first postback already says who registered and who paid.
Modelled attribution
The platforms fill gaps with modelled conversions. Useful for spotting trends inside one network. Never the only number we use to move budget between networks.
We Test Angles, Not Just Videos
Changing the colour of a button is not a test. Changing why someone should download is. Every quarter we run angles against each other in a bracket, and the winner sets the brief for the next month of production.
Chat, events and groups in one app.
Your Saturday crew is already here.
I moved cities and found my people in a week.
Join free this weekend.
Cheaper registrations and better Day-7.
Strong clicks, weaker retention.
Judged on cost per profile complete and Day-7 retention.
One video can fail for reasons that have nothing to do with the idea. Three gives the angle a fair chance.
Click-through rate picks curious scrollers. We pick the angle that brings people who stay.
Long enough to get past the learning phase, short enough to keep production moving.
Sample bracket for a fictional social app, to show how a test runs.
Your Store Page Is Half the Ad
Paid traffic lands on your App Store or Play Store page. If that page converts badly, every channel pays more per install. We fix the listing alongside the campaigns.
- Keyword in the title. The words people search for go where the stores weigh them most.
- Subtitle sells the outcome. What the user gets, not what the app is.
- One benefit per screenshot. Big captions, readable at thumbnail size.
- Custom pages per ad set. Custom product pages on iOS and custom store listings on Android, so the page matches the ad.
How an Indian Social App Cut CPI by 48% While Improving Day-30 Retention
The problemCaarisma was paying a high cost per install, and too few of those users were still around at Day 30.
What we changed- Restructured campaigns to optimise for registration and profile completion events instead of installs.
- Implemented Firebase attribution, so every platform was measured the same way.
- Shifted creative to community-belonging angles.
Our App Marketing Launch Process
Four updates, installed in order. Nothing scales until the one above it is done.
MMP & Attribution Setup
Week 1SDK checked on both stores, an event list agreed with your product team, SKAN conversion values mapped, postbacks sent to every network and test installs run on real devices. No spend until the numbers match.
Creative Asset Production
Weeks 1-2Six to ten videos per platform for the first round, in vertical, square and landscape where needed, plus store screenshots and custom product pages that match the ads.
Campaign Launch & Event Optimisation
Weeks 2-3Separate Android and iOS campaigns go live optimising for registration. We check cost per event daily through the learning phase and leave settings alone unless something is clearly broken.
Scale Quality Users, Suppress Churn
Week 3 onwardsBudget moves to the cohorts with the best Day-7 and Day-30 numbers. Existing users are excluded from acquisition, low-retention segments are cut, and app retargeting brings lapsed users back.
From MMP setup to scaled campaigns in 2-3 weeks.
When Does a User Pay for Themselves?
The question behind every budget decision. Here is a sample subscription app: $38 to acquire a paying user, $9.99 a month, about $8.49 after the store’s cut. The battery fills as each cohort earns back what it cost.
Sample numbers for illustration. We build this curve from your own MMP and store revenue data before we scale.
App marketing services in detail
Channels we pair with app UA
App Marketing Agency Questions, Answered
What founders and growth leads ask us before we touch their ad accounts or SDK.
Ask us on a free auditWhich MMP do you work with?
All three of the common ones. Firebase is free and works well for smaller budgets, especially Android-heavy apps. Adjust or AppsFlyer make more sense once you need LTV reporting, cohort analysis, fraud protection and clean SKAN handling across several networks. If you already have one, we work with it.
How is app marketing affected by iOS ATT?
Only users who allow tracking can be measured one by one. For everyone else we rely on SKAdNetwork (and Apple’s newer AdAttributionKit) alongside MMP postbacks and modelled attribution. In practice that means a careful conversion value schema, patience with delayed data, and judging iOS on cohorts rather than on yesterday’s numbers.
Do you run Android and iOS simultaneously?
Yes, but always as separate campaigns per store. Costs, measurement and user behaviour differ too much between the two to share a budget. Mixing them usually means the cheaper store eats the whole budget.
Can you help with App Store Optimisation?
Yes. We work on titles, subtitles, keywords, screenshots, preview videos and custom product pages for the App Store and Google Play. Paid and organic share the same store page, so improving its conversion rate lowers the cost of every paid install too.
Should we optimise for installs or in-app events?
In-app events, as soon as you have the volume. Install optimisation finds people who install, and many of them never open the app twice. We usually start on registration and move to deeper events such as profile completion or first purchase once they fire often enough.
How long before we see results?
Setup and the first campaigns take two to three weeks. Cost per install and cost per registration settle in the following two to four weeks. Day-30 retention, by definition, takes a month after each cohort installs, so the honest read on quality comes at around week eight.
What creative do you need from us?
Access to the app, a brand kit and screen recordings of the key flows are enough to start. Our creative team produces the ad videos, statics and store screenshots. If you have creators or community members willing to film, those clips usually outperform polished studio work.
Do you work with subscription apps and games?
We mostly work with consumer apps: social, community, subscription and marketplace apps. The same approach applies to subscriptions, with trial starts and paid conversions as the key events. For mid-core and hard-core games, a specialist games UA team is usually a better fit, and we will tell you so.
How do you measure Apple Search Ads?
Apple Search Ads attribution comes through Apple’s AdServices framework and does not need the ATT prompt, so most MMPs report it reliably. We judge each keyword group on registrations and revenue in the MMP, not on taps or installs in the Apple dashboard.
Can you work with our existing ad accounts and MMP?
Yes, and we prefer to. Account history and pixel or SDK data help the algorithms learn faster. You keep ownership of every account, and we audit the existing setup before changing anything.
Is there a minimum contract?
Yes, three months. Day-30 retention needs at least a couple of full cohorts before anyone can judge it fairly. After that it runs month to month.