How does AppLovin find buyers without user-level targeting?
Our answer
AppLovin's model picks who sees each ad from behavior across its game network and the purchase data you send. You set only countries, goal, audience strategy and creative, so feed it clean purchase data and say who it's for.
- Confidence
- Medium
- Strongest proof
- 8 AppLovin rules
- 2 practitioners, 10 opinions or claims
- Sources
- 8 sources
- 20 statements
- A
SPGDNAD
New here? Take the 1-minute tour
Why this is the answer
3 reasons and our judgmentYou get almost no audience controls. AppLovin says its model primarily handles targeting, and its web API offers only country, US state and metro fields. Will Frank of AS Beauty told AdExchanger there are no audience controls.
The model runs on the data you share. AppLovin's CEO says advertisers must share their data for the model to work. AppLovin says Prospecting and Discovery need your order history to know who already bought.
It predicts buyers from behavior across AppLovin's game network, and your creative is the main way to steer whom it reaches. Jetfuel Agency, Sweat Pants Agency and GabGrowth, an investor newsletter, all describe the model this way.
Our judgment
Treat the model as the targeting and your data as its input, so connect the pixel, Conversion API and order history before launch. That follows AppLovin's FAQ and launch checklist, its CEO's statement on sharing data and Sweat Pants Agency's account of the model learning from purchase events. Let the creative name the audience, going by Sweat Pants Agency's view of creative as the primary lever for audience selection, Apt Digital's advice to say who the product is for in the hook and AppLovin's finding that about 30% of its top web videos define a group the product is for. Read the lack of user-level targeting as no audience controls for advertisers, not as no user data in the model. That reconciles the description by Common Thread Collective (CTC) of contextual signals with AppLovin's filing on personal information and its CEO's account of users' transactional behavior.
Our judgment, built from AppLovin, Sweat Pants Agency, Apt Digital and Common Thread Collective: our reading of these sources, not a finding from any one of them.
What the sources say
20 statements from 8 sourcesAppLovin says its ads run in mobile games such as Candy Crush, its model handles targeting, and it claims 150 million+ US daily users. It suggests leaving all metros and states selected.
“Generally, leave all metros and states selected and let AppLovin's optimizer shift spend accordingly.”
AppLovin's web campaign API targets by country, plus US states or US metro areas; no app, age or audience field appears. Budgets and goals can be set per country.
“The country targets for the campaign”
AppLovin says purchase campaigns can use a Universal, Prospecting or Discovery audience strategy, and you can run several at once. Prospecting and Discovery need Shopify connection or uploaded purchase history.
“Discovery campaigns optimize for purchases from users who never visited your site before.”
AppLovin's FAQ says to connect the Shopify app or upload past purchase data for prospecting, and to start with one campaign so the model gets more data. More creative volume and variety also help.
“Start with a single campaign.”
AppLovin highly recommends sharing order history to optimize prospecting, either through its Shopify app or a CSV upload, subject to your own privacy notices and consents.
“To optimize prospecting campaigns (which acquire new customers), AppLovin highly recommends that you provide AppLovin with order history data.”
AppLovin recommends running its Conversion API alongside the AppLovin Pixel, sending the same dedupe_id for matching pixel and server events, in batches of up to 100.
“The same pixel event and server-to-server event should have the same dedupe_id”
AppLovin's launch checklist says to target all available regions outside Western Europe and group several countries in one campaign for faster learning. It gives no reason for excluding Western Europe.
“Drive scale and efficiency by targeting all available regions outside of Western Europe.”
AppLovin's review of the top 1,000 web advertiser videos by spend found 95% had prominent captions, 85% voiceover, and most a UGC style. It advises uploading many portrait videos, including some over 30 seconds.
“You do not need to pause creatives. AppLovin's optimizer automatically reduces spend on creatives that don't perform.”
AppLovin's privacy rules make the advertiser fully responsible for privacy law compliance and consent, require listing AppLovin in the privacy policy, and bar sharing sensitive data or custom events with it.
“you are fully responsible for complying with applicable privacy laws”
Apt Digital, an AppLovin partner agency, says to adapt Meta winners rather than upload them: stretch a 15 to 20 second ad to about 45 seconds with a demo, proof and offer, and add voiceover.
“A 15 to 20 second Meta winner can become a 45 second AppLovin cut by adding a longer demo, a proof block and a stated offer.”
An investor newsletter describes AppLovin's AXON as a real-time model that predicts each user's chance of a purchase or install and bids on it, using behavior signal from over 1 billion devices.
“AXON is a real-time machine-learning system built around reinforcement learning.”
AdExchanger reports that AppLovin campaigns are simple to set up, with no audience or keyword controls, just Discovery or Prospecting, and buyers praise its account service compared with Google and Meta.
“There are no audience controls, no keyword strategies, no platform bells and whistles of any kind”
AdExchanger reports AppLovin gives advertisers little transparency, with no audience or app-level detail, only OS, time of day and ad unit type, and serves ads only in mobile game apps.
“AppLovin serves ads in mobile game apps. That's it.”
AppLovin's CEO says advertisers must share their data with its model whatever their spend, and it wants more mid-market customers who give visibility into users' transactional behavior. Its own account of its model.
“They have to share their data with us to tap into the deep learning model that we have to get output, good advertising results out of our system.”
AppLovin's 10-Q for the quarter ended June 30, 2026 says it processes personal data, shares some with demand partners, and lists Apple and Google policy changes as risks. It does not say which signals its model uses.
“Axon AI, our advertising recommendation system”
On its Q2 2026 call, AppLovin's CEO said consumer advertiser spend hit a record 28% above Q4 2025, but its ecommerce model is still early and small shops spending little are less likely to hit their goals.
“mid-market is a sweet spot right now.”
Pilothouse, which sells AppLovin management, says AppLovin's game network reaches all ages and to let it run at first, adding age targeting if delivery drifts. AppLovin's API docs and FAQ don't mention that setting.
“at the start, we would suggest it to just let it do its thing”
Jetfuel says AppLovin's Axon is a value bidder with no manual model controls that needs around ten conversions a day before its lookalike model fully engages, and advises launching with Universal audience.
“Axon is a value-based bidder that needs around ten conversions per day before its lookalike model fully engages”
Sweat Pants Agency, which runs AppLovin tests, says AppLovin offers no demographic or interest targeting and its model learns from purchase events, so creative is the main lever for reaching a persona.
“Creative becomes the primary lever for audience selection.”
CTC, an agency that runs AppLovin for clients, says AppLovin's Axon engine targets with contextual signals, not user-level identifiers, which sidesteps iOS privacy limits. AppLovin's own pages do not use that wording.
“Axon is AppLovin's proprietary AI advertising engine. It powers ad targeting, creative optimization, and bid decisions”
In the full answer
For readers who sign in- What to do this week3 steps
- The numbers behind it3 figures from cited sources
- Where each source standsA chart placing each source on the question
- The strongest disagreementThe best case against the answer, from a cited source
- How sure we areWhat its confidence rests on, and what would change the answer