Where Does Google Get the Data Behind Its Audience Segments? 🕵️
Google’s audience segments are one of programmatic advertising’s most fascinating mysteries. We have already asked why some segments are so large and which signals Google might use. But another question remains: where does the underlying user data come from? 🤔
1. Where might Google obtain signals for audience segments?
There are two broad categories to consider. First, Google’s own ecosystem: Search, YouTube, Gmail, Gemini, Maps and other services. Second, third-party apps that integrate Google technology.
We have already explored the unusual size of some Google audience segments and the signals Google may use to build them. This time, we wanted to examine the presence of Google trackers in other companies’ apps.
One important distinction: the presence of a Google SDK or tracking endpoint does not establish that a particular event is used for advertising personalisation or audience modelling (the purpose, consent conditions and data flows may differ by product and configuration).
2. How did we investigate Google tracking in French apps?
We selected a range of popular apps in France and used DuckDuckGo’s App Tracking Protection, which identifies tracker activity observed while Android apps are in use.
Our checks have three important limitations:
- 📱 Android only: the tool does not provide an equivalent observation of iOS behaviour.
- 🖥️ No server-side visibility: it cannot detect every data flow occurring between an app’s servers and third parties.
- 👆 Only observed activity: it reports trackers encountered in the sections and sessions we visited; a negative result is not evidence that an app never communicates with Google.
Our source post does not document the exact test dates, device configuration, app versions or session lengths, so the findings should be read as a snapshot of our observations, not as a reproducible market-wide audit.
3. Which apps showed Google tracking activity?
Here are the results from our original checks, grouped by category. ✅ means App Tracking Protection detected Google tracking during the test; ❌ means no Google tracker was detected in the observed session.
| Category | Apps and observed results |
|---|---|
| Audio | ✅ CastBox, ✅ Deezer, ✅ France Info, ✅ Podcast Addict, ✅ TuneIn Radio, ✅ Spotify |
| Banking | ✅ BoursoBank, ✅ Caisse d’Epargne, ❌ Crédit Mutuel, ✅ Macif, ✅ Qonto |
| ✅ La Poste, ✅ Mail Orange, ✅ Yahoo | |
| News | ✅ 20 Minutes, ✅ L’Equipe, ✅ Le Monde, ✅ Le Figaro, ✅ Le Parisien, ✅ Les Echos, ✅ Ouest France, ✅ ProgrammeTV, ✅ SudOuest |
| Retail | ❌ Action, ✅ Auchan, ✅ Carrefour, ✅ LeBonCoin, ❌ Leclerc, ❌ Lidl +, ✅ Magasin U, ✅ Temu, ✅ Vinted |
| Video | ❌ Arte, ❌ FranceTV, ✅ HBO, ✅ M6+, ✅ myCanal, ❌ Netflix, ❌ TF1+ |
| Travel | ❌ Airbnb, ✅ AirFrance, ✅ Citymapper, ✅ EasyJet, ❌ edreams, ❌ Flixbus, ✅ Getaround, ✅ Kayak, ✅ Pages jaunes, ✅ Ryanair, ✅ Skyscanner, ✅ SNCF, ✅ Trainline, ✅ Transavia |
That is 42 detections across 53 apps checked in this informal sample. These results are neither an estimate of Google’s overall reach nor a measurement of how many users’ data feed its audience segments.
What caught our eye? Google tracking was detected across many categories: music and podcasts, banking, email, news, retail, video and travel. The pattern suggests that Google technology is widely integrated into the mobile-app ecosystem we sampled.
4. Does a Google tracker mean an app enriches Google Ads audience segments?
No—not necessarily. Apps may use Google services such as AdMob, Google advertising infrastructure, Firebase or Analytics for different purposes. Detecting a Google-associated request alone does not reveal which data fields were transferred, the legal basis, or whether those data were used to build advertising audiences.
But it does illustrate why the question is worth asking. The technical integrations are present in many places, beyond Google’s own apps. The precise flow of personal data and its use for audience modelling would require additional evidence, such as SDK documentation, network-payload analysis and product-specific privacy disclosures.
In other words: the ingredients may be visible, but that does not tell us the recipe. 😉
5. What should programmatic media traders take from this?
For media traders and agencies, the practical lesson is to ask more precise questions about data provenance, consent, signal availability and audience construction. “Google audience segment” is a product label; it is not, on its own, a full description of the underlying signals.
This is also a reminder that data from publishers and apps can create value across the wider AdTech ecosystem. We have written about first-party and 1.5-party data and the challenge of activating publisher data through the programmatic supply chain.
Want to repeat our experiment? Install DuckDuckGo App Tracking Protection on a compatible Android device, use the apps you wish to investigate, and document the precise sessions and conditions. Treat the results as observations, not proof of downstream audience enrichment.
6. How does GreyFox Ads approach transparency and audience targeting?
Here in France, GreyFox Ads believes that a quality Open Web can deliver meaningful effectiveness for advertisers. Our Master Curated Deal is designed to make programmatic buying easier to activate and understand, with curated inventory, quality controls, transparent supply paths and appropriate audience strategies across cookie-based and cookieless environments (subject to the applicable privacy requirements).
And when we investigate a mystery like Google’s audience data, we prefer to show what we actually observed—and what remains unknown. 🕵️
Original LinkedIn post: Read our original investigation.
Earlier investigations: Audience segment sizes · Signals collected.
Looking for a quality programmatic partner in France?
Want to make your Open Web display and video buying simpler, more effective and more transparent?
Contact GreyFox AdsDiscover the Master Deal