Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Meta Ads

Meta Ads for App Installs

A recommended Meta Ads setup for app installs: the App promotion objective, SDK or partner events, install versus app event optimisation and in-app creative.
Objective guideintermediate7 min readUpdated

Why app installs change the Meta Ads setup

Website campaigns rely on a pixel watching a browser. An app campaign cannot: the result happens inside an app, after a trip through a store Meta does not control, on a phone whose operating system may restrict what any advertiser learns about the person. The signal has to come from code inside the app, either Meta's SDK or a measurement partner that passes events on, and on iOS much of it arrives late and in aggregate. At the same time, installs themselves are abundant. Delivery is very good at finding people who will tap install, which makes cost per install look healthy while most of those people never get past the first screen. The setup is about what Meta learns from after the install and how to give it enough of that event to work with. The app marketing hub places paid acquisition beside store optimisation and retention, and the app install campaigns guide compares Meta with the other networks.

SettingRecommended for app installsWhy it differs from other objectivesFeature page
ObjectiveApp promotionThe only objective whose results are installs and in-app eventsCampaign objectives
EventsInstall plus an activation event through the SDK or a measurement partner, tested before launchWebsite objectives read a pixel; an app needs code inside the productApp attribution and SKAdNetwork
Optimisation goalApp installs at launch; an app event once its volume lets the ad set exit learningFew other objectives have such a gap between the cheap result and the useful oneLearning phase
Store splitOne campaign per store listingiOS and Android report under different privacy rulesApp attribution and SKAdNetwork
CreativeVertical video of the app screen in use, captions for silent viewingThe product is the interface, so the ad has to show itAd formats and placements

Events wired and tested before spend

An app campaign without working events can still buy installs, and it will be judged on them because nothing else is visible. Before launch, decide which in-app actions matter, typically registration, an activation step and a purchase or subscription, and send them through the Meta SDK or through the measurement partner the app already uses. Firebase and similar SDKs can supply the app analytics; the partner, where there is one, should be the single source for attribution. Fire each event from a test device and confirm it in Events Manager and in the partner's dashboard. The app attribution guide explains how the partner and Apple's SKAdNetwork fit together.

Stepping from installs to an app event

Install optimisation finds people likely to install. Event optimisation finds people likely to do something after installing, which costs more per install and less per person who stays. The limit is volume: Meta's help centre describes a learning phase that needs a flow of the optimisation event before delivery settles, and the learning phase guide shows how rare events leave ad sets learning limited. Choose an activation event early in the journey that most retained users perform, let it accumulate while optimising for installs, then switch once the volume at your budget is enough. A deeper event such as a subscription comes later, if ever.

A campaign for each store

Each App promotion campaign points at one store listing, and the two platforms report differently. On Android the install referrer carries campaign details into the app; on iOS, for people who have not allowed tracking, results come through SKAdNetwork postbacks that are delayed and coarse. Mixing the two hides which platform is working and complicates event optimisation on iOS, where event choices need to be simpler. Give each store its own campaign, budget and event plan.

Creative that shows the app working

The decision to install is made by watching the app do its job. Open on the moment of value, show the real interface, caption everything for silent autoplay and build for vertical placements such as Stories and Reels, where app ads sit naturally. The ad formats and placements guide lists vertical video of the app in use as the starting format for app promotion. Refresh often: an audience that keeps seeing the same screen recording stops responding as Frequency builds.

Lapsed installers kept apart

People who installed and went quiet are a different audience with a different message: what they missed, what changed, a reason to open again. Retarget them in their own campaign so their re-engagement does not mix with acquisition results, and exclude them from acquisition so new-user budget reaches new users.

A worked app installs setup

Illustrative example — not a real result

Pantrywise, an invented meal-planning app

Pantrywise is free to install, with a subscription after a trial. Its team knows that users who create a first weekly plan in their opening week are far more likely to subscribe. The first Android campaign optimises for installs at £70 a day with landscape trailer videos.

SettingFirst monthFrom week seven
EventsInstall onlyInstall, registration, "first plan created" via the partner, tested on devices
Optimisation goalApp installsApp event: first plan created
Store splitAndroid onlyAndroid and iOS in separate campaigns, iOS on installs while its event volume builds
CreativeLandscape trailerVertical screen recordings: scanning the fridge, a plan appearing, a shopping list
RetargetingNoneLapsed installers in their own campaign at £10 a day

In the first month Android spend of £2,170 bought 2,890 installs at £0.75 each, but only 260 users created a first plan, £8.35 per plan. Over weeks three to six the event reached around 70 a week, enough to switch. In the month after the switch the same spend produced 1,240 installs at £1.75 each and 470 first plans at £4.62 each, and the trial-to-subscription rate among those users rose from 11% to 16%.

The install count fell by more than half and the business improved. Delivery had been finding people who tap install; once it learned from the plan event, it found people who use the product, and the creative showing the plan being built set the right expectation before the download.

Measuring app installs in Meta Ads

Ads Manager reports App installs and Cost per app install as the results of an install-optimised ad set, and App events with Cost per app event for event optimisation. Watch the Learning status column after any switch of goal, since a new event restarts learning and the first days are unstable. Set the results reported by the measurement partner beside Meta's own figures: the partner deduplicates across networks and is the better record, while Meta's numbers drive in-platform decisions. On iOS expect delay and aggregation in both.

The business measures are cost per activated user and the value of a cohort over time, its lifetime value (LTV) against acquisition cost (CAC). The CPA calculator turns install cost and activation rate into cost per activated user (CPA), and the marketing budget calculator sizes spend against that target. Check event health and learning daily after a change, cost per app event weekly, and cohort retention monthly. The growth loops framework for app installs shows how paid installs can feed referral and content loops rather than standing alone.

Where the app setup goes wrong

Switching to a deep event that happens rarely leaves the ad set learning limited and spending unevenly. Changing the goal and the creative at once removes any chance of knowing which helped. Treating Meta's install figure as the record leads to double counting across networks. The app install campaign launch checklist settles which measurement partner holds the record before any network goes live. And on iOS, a complex event plan built for Android does not translate: keep iOS events simple and judge them over longer periods.

Across networks, paid social for app installs covers the channel-level tactics and a budget that scales against cohort payback rather than cost per install.

Common mistakes

  • Launching before events fire. The campaign can only be judged on installs.
  • Staying on installs indefinitely. Cheap installs with no activation are not users.
  • One campaign for both stores. Results and event optimisation blur together.
  • Landscape trailers in vertical placements. Show the app screen, full height.
  • Mixing re-engagement with acquisition. Lapsed users flatter acquisition figures.

Frequently asked questions

Should a new app's Meta campaign optimise for installs or app events?

Start with installs only if the in-app event you care about is still too rare for an ad set to learn from. Pick an early activation event that predicts retention, such as completing onboarding or creating a first item, confirm it reports correctly, and move to it once it occurs often enough at your budget. Installs are a cheap result that says little about whether people use the app.

Why do Meta and our measurement partner report different install numbers?

They apply different rules. Meta credits installs to its own ads using its attribution windows, while a measurement partner applies one set of rules across every network and removes installs it attributes elsewhere. On iOS, privacy rules limit what either can see and results arrive delayed and aggregated. Pick the partner's figure as the record and use Meta's for in-platform decisions.

Do iOS and Android need separate Meta app campaigns?

Yes, in practice. Each campaign promotes one store listing, and the reporting rules differ: Android installs can be attributed through the install referrer, while iOS reporting for people who have not allowed tracking works through Apple's limited, delayed postbacks. Separate campaigns keep budgets, events and results readable.

Terms used on this page

Tools to try

Frameworks

Where this page is used

Also related

Sources

  1. Meta Business Help Centre — Meta Business Help Centre
  2. About the learning phase — Meta Business Help Centre
  3. Firebase documentation — Google
  4. SKAdNetwork — Apple Developer

Related