Lookalike from app users

Quick Answer

App user lookalikes are some of the most powerful seeds Meta offers — but only if your Facebook SDK is properly installed and you are firing meaningful events (purchases, subscriptions, level completions, key actions). Build the lookalike from a high-value event segment (e.g., users who completed onboarding AND made a purchase) rather than all installs. Expect 30% to 50% lower CPI compared to broad install campaigns.

Why app user lookalikes outperform install lookalikes

Most app marketers make the same mistake: they build a lookalike from "all app installs." The problem is that the install event includes everyone who tapped Install — including the 60%+ who never opened the app, never registered, never paid. Meta is modelling against churn.

A segmented app user lookalike is built from people who did the thing that makes you money. That signal is dramatically stronger.

Step-by-step setup

1. Install the Facebook SDK or use a mobile measurement partner like AppsFlyer, Adjust, or Singular. Your engineering team needs to fire standard events: app install, registration, level completion, in-app purchase.
2. Confirm events are flowing. Go to Events Manager and check that your app events are arriving. Dedupe carefully if you use both SDK and MMP.
3. Create a custom audience from a high-value event. Audiences > Create > Custom Audience > App Activity. Pick the event (e.g., purchase, subscription start, level 5 completion). Use a 90-day window.
4. Layer events for stronger seeds. Create custom audiences like "installed AND completed onboarding AND purchased" using the multi-event filter.
5. Build a 1% lookalike from the high-value custom audience. Name it LAL_1%_AppPurchasers_v1.
6. Run a separate ad set with App Promotion as the objective and the lookalike as the audience. Optimise for the same high-value event you seeded from.

Real examples

Mobile fitness app, $9.99/month subscription. Built a lookalike from "users who started a free trial." CPI was $4.20, trial-to-paid was 18%. Switched seed to "users who completed onboarding AND started a trial AND opened the app on day 3." New lookalike CPI was $5.10 (higher), but trial-to-paid jumped to 32%. Net cost per paying subscriber dropped 38%.

Mobile game. Seed: "all installs." CPI: $0.80, Day 7 retention: 8%. Switched to "installs who reached level 5." CPI: $1.40, Day 7 retention: 22%. Day 30 LTV more than tripled.

Travel app. Seed: "users who completed a booking." CPI was 2x the install lookalike but acquired users had 4x the booking rate. Net ROI tripled.

Best app events to seed from

• Purchase or subscription start — strongest signal
• Onboarding completion + first action — strong signal, scales faster than purchase
• Repeat sessions (Day 3, Day 7) — good signal for retention-focused apps
• Key milestone reached — good for games and progression apps
• Trial start — good for freemium SaaS apps

What kills app user lookalikes

• Seeding from "all installs" (too noisy)
• Broken or duplicated SDK events (Meta cannot trust the signal)
• Tiny seeds (under 500 high-value users)
• Mixing iOS and Android into one seed if behaviour differs sharply
• Ignoring iOS 14+ AAA (Aggregated Event Measurement) limits — you only get 8 events

FAQs

Do I need the Facebook SDK or is an MMP enough?
Either works. If you use an MMP, configure it to send events to Meta via the Conversions API or the SDK postback. Avoid double-counting.

How are iOS 14+ changes affecting app lookalikes?
Significantly. Apple's ATT prompt limits how Meta can track iOS users. You only get 8 events per app via Aggregated Event Measurement. Pick your 8 events carefully and prioritise the ones that drive value.

Should I build separate lookalikes for iOS and Android?
Yes, especially if your app behaves differently on each platform. Different audiences also helps with creative testing.

Can I use the Conversions API for app events?
Yes. CAPI for app sends events server-to-server and improves event matching, especially on iOS where SDK signals are limited.

How often should I refresh app user lookalikes?
Meta refreshes automatically every few days. Manual refresh is only needed if you change the seed source.

Make app store and ad creative work harder

App install ads compete on creative. Pix-Vu tests app store screenshots, video previews, and ad creatives against real audience reactions so your install rate matches the quality of your audience targeting. Try Pix-Vu free at https://pix-vu.com.