Meta ads for app installs:

To run Meta ads for app installs, set up the Meta SDK or your MMP with real in-app events first, then run one app promotion campaign in one country, on one platform, with a small batch of creatives. Set the weekly budget at about 50 times your expected cost per event so it can leave the learning phase, and judge it on cost per activated user, not cost per install.

This is a first test plan for an app that hasn’t run Meta ads before, or ran them without clear tracking. It uses Meta’s and Apple’s own help pages for the setup facts, and your own numbers for everything else. If you’re not sure you’re ready to spend at all, read when to start paid ads first.

The short version

• Install the Meta SDK (or connect your MMP) and send real in-app events before you spend anything.
• Start with one app promotion campaign using the Advantage+ setup, aimed at one country and one platform.
• Size the budget from your own cost per event, so the campaign can get out of Meta’s learning phase.
• Test a small batch of creatives at a time. Creative is where most of the difference comes from.
• Judge the test on cost per activated user and payback, not cost per install.

Before you spend: tracking

Meta can only optimise for what it can see. Its help page on the app promotion objective lists three things you need:

1. An app registered on Meta for Developers.
2. An ad account linked to that app.
3. The Meta SDK (Facebook SDK) and App Events. You need these to optimise for installs, app events or value, and to measure them.

Without the first two, Meta says your ads will optimise for link clicks instead of installs. That’s the most common reason a first campaign “gets clicks but no users”.

SDK or MMP. You can send events with the Meta SDK, through Meta’s Conversions API or App Events API, or through a mobile measurement partner (MMP) such as the ones you might already use for other networks. If you use an MMP, Meta says to check that it supports Apple’s SKAdNetwork before you rely on it for iOS. (Meta: prepare for SKAdNetwork) An MMP is most useful once you run several ad networks and need one view across them. With Meta alone, the SDK is often enough to start.

The events that matter. Send at least: install, completed registration, your activation event (the first moment a user gets real value), and purchase or subscription start with its value. Our guide on how to track app installs walks through the setup.

iOS limits. On iOS, apps must ask permission through Apple’s App Tracking Transparency framework before tracking people across apps and websites. For ad measurement, Apple provides AdAttributionKit and the older SKAdNetwork. Their postbacks don’t include user or device level data, the level of detail depends on a privacy threshold Apple sets, and results arrive in conversion windows rather than in real time. In Meta Ads Manager, an iOS 14+ app campaign uses either Meta’s Aggregated Event Measurement or Apple’s SKAdNetwork as its attribution method. (Meta: iOS 14.5 campaign considerations) In practice, expect iOS numbers to be less detailed and slower than Android, and check them against your own sign-up and revenue data.

Which campaign type to start with

Use the app promotion objective. Meta notes it used to be called app installs, and it can optimise for installs, app events or value. (Meta: app promotion objective)

Within it, start with an Advantage+ app campaign. Meta describes it as using its AI to optimise bidding, audiences and placements, with simpler inputs: you choose country, language and app store or operating system, and Meta finds the people. (Meta: Advantage+ app campaigns) For iOS 14.5 and later, Meta says only the Advantage+ setup is available for these campaigns. (Meta: iOS 14.5 campaign considerations)

Keep the first test simple: one campaign, one country, one platform. Meta’s own iOS guidance suggests testing creative on campaigns for Android users first, then bringing the best ones to your iOS 14+ campaigns.

Budget: size the first test from your own numbers

Meta’s delivery system needs data to settle. An ad set usually leaves the learning phase after about 50 results in the week after its last significant edit, and it’s marked “learning limited” when it’s unlikely to get there. (Meta: about the learning phase, Meta: learning limited)

So work the budget backwards:

1. Pick the event you’ll optimise for (install, registration or an in-app event).
2. Estimate what one of those events will cost. Use your past ads, another network, or a careful guess you’ll correct after a few days.
3. Weekly test budget = about 50 x your expected cost per optimisation event. Divide by 7 for a daily budget.

If that number is more than you can spend, don’t cut it until the campaign can’t learn. Optimise for a more frequent event instead, as Meta suggests for learning limited ad sets, and track the deeper events in your own analytics. Then decide how long you’ll run the test before you look at the results, and don’t edit it while it’s learning, because edits restart learning.

Creative: what to test

Creative is the main lever you control, especially with Advantage+, where targeting is mostly left to Meta.

• Hooks. The opening decides if anyone keeps watching. Test different opening lines and first shots before you test anything else.
• UGC-style video. A real person using the app on their phone, talking plainly. Our guide to UGC ads for apps covers how to brief and test it.
• Statics. One clear screen and one benefit. They’re cheap to make and good for testing messages fast.

Test in small batches. Meta’s iOS guidance suggests up to six creative options in an ad set. Change one thing at a time between batches (the hook, the person, the format), so you know what made the difference. If you also run TikTok, our TikTok marketing for apps guide covers the creative style that platform expects.

Targeting: broad vs interests

Advantage+ app campaigns don’t support interest, behaviour, demographic or lookalike targeting. You set country, language and operating system, and you can exclude existing users with custom audience exclusions. (Meta: Advantage+ app campaigns) So for a first test, go broad and let the creative do the targeting: an ad that speaks to one type of user will mostly reach that type of user.

What to measure

An install is the start, not the result. Follow each campaign down three steps:

1. Install. Did the ad bring people into the app?
2. Activation. Did they reach the moment of real value, like finishing onboarding or completing the first core action?
3. Paid. Did they start a trial, subscribe or buy?

The number to judge the test on is cost per activated user: spend divided by users who reached activation. Then compare it with what an activated user is worth to you, and check how long it takes to earn the cost back with our CAC payback calculator. Two campaigns can have the same cost per install and very different cost per activated user.

When to kill, keep or scale a test

Set these rules before launch, in your own numbers, so you don’t decide on feel.

• Kill a creative or campaign that has spent enough to give a fair read and still has cost per activated user well above what a user is worth to you.
• Keep testing when cost per activated user is close to your target. Try new hooks before you touch the budget.
• Scale when cost per activated user is below target and paid conversion holds up. Raise the budget in steps rather than all at once, since Meta’s guidance is to avoid frequent budget changes that can restart learning.

Common mistakes

• Launching before the app is registered and the ad account is linked, so Meta optimises for clicks.
• Judging the test on cost per install and ignoring what happens after.
• Running many small ad sets that each stay learning limited.
• Editing the campaign every day while it’s still learning.
• Testing ten creatives that are all the same idea.
• Trusting iOS numbers in Ads Manager without checking them against your own data.

FAQ

How much should I spend on my first Meta app install campaign?

Work it out from your own numbers. Estimate the cost of the event you’ll optimise for and multiply by about 50, since Meta says ad sets usually leave the learning phase after about 50 results in a week. If that’s too much, optimise for a more frequent event.

Do Meta install ads work on iOS after ATT?

Yes, but measurement is more limited. iOS 14+ campaigns use Meta’s Aggregated Event Measurement or Apple’s SKAdNetwork for attribution, and Apple’s postbacks are delayed and less detailed. Check results against your own sign-up and revenue data.

Should I optimise for installs or in-app events?

Optimise for the deepest event you can get about 50 of a week. For most new apps that starts as installs or registrations, then moves to activation or purchase once volume allows. Always measure the deeper events, whatever you optimise for.

When should a startup start paid ads?

When tracking works, your activation step is clear, and you know roughly what a user is worth, so you can tell if an ad paid back. Our guide on when to start paid ads covers the signs.

Want a channel you don’t pay per click for?

Ignura is an AI-native agency that automates your GEO and SEO. You approve what goes out. We don’t run ads. We work on the pages people find when they search on Google and in AI answers: AI search, technical SEO and content. Plans are on the pricing page.

Book your free Growth Check