Advertising Deployments
Deployment evidence
Advertising PET deployments are often platform-controlled. They may be production systems, but the privacy and utility evidence can still be difficult for outsiders to inspect.
Production Platform Deployments
Google Ads Data Hub
| Field | Entry |
|---|---|
| Organization / project | Google Ads Data Hub |
| Domain | Advertising |
| Problem | Let advertisers run customized analysis on Google advertising data while enforcing privacy checks on query outputs. |
| PETs used | Data clean room, privacy checks, controlled query environment |
| Deployment maturity | Production |
| Source quality | Primary / official vendor documentation |
| What worked | Ads Data Hub is a documented Google product with UI/API workflows and privacy checks on query results. |
| Challenges | It is platform-governed; users must learn the operating model, and privacy depends on query rules, access, and output checks. |
| Lessons for builders | Clean rooms are as much governance and workflow as technology. Query approval, thresholds, and output review are core product features, not add-ons. |
| Source | Google Ads Data Hub documentation and Ads Data Hub introduction |
(Evidence: Deployment-backed. Source quality: Primary / official vendor documentation. Reviewed 2026-06-17 — strong evidence the product and controls exist; weaker evidence for independent privacy or utility outcomes.)
Apple SKAdNetwork / AdAttributionKit
| Field | Entry |
|---|---|
| Organization / project | Apple SKAdNetwork and AdAttributionKit |
| Domain | Advertising |
| Problem | Measure app-install and post-install campaign performance without exposing user-level identifiers in the same way as traditional attribution. |
| PETs used | Privacy-preserving attribution framework, aggregation/delay/thresholding concepts |
| Deployment maturity | Production |
| Source quality | Primary / official vendor documentation |
| What worked | SKAdNetwork is part of Apple's developer ecosystem and is used for iOS app attribution workflows. |
| Challenges | Measurement is less granular; conversion-value design, delays, thresholds, and interoperability create operational pain for advertisers. |
| Lessons for builders | Privacy-preserving attribution changes the optimization workflow, not only the data transport. Expect utility loss and new measurement practices. |
| Source | Apple Developer: SKAdNetwork and Apple Ads attribution overview |
(Evidence: Deployment-backed. Source quality: Primary / official vendor documentation. Reviewed 2026-06-17 — production framework evidence is strong; exact measurement utility depends on campaign setup and current Apple platform behavior.)
Chrome Privacy Sandbox Attribution Reporting
| Field | Entry |
|---|---|
| Organization / project | Chrome Privacy Sandbox Attribution Reporting API |
| Domain | Advertising |
| Problem | Enable conversion measurement without third-party cookies. |
| PETs used | Browser-mediated attribution, event-level and aggregate reports, noise for summary reports |
| Deployment maturity | Production / evolving platform feature |
| Source quality | Primary / official vendor documentation plus platform documentation |
| What worked | The API is documented for web developers and supports attribution flows with privacy controls. |
| Challenges | Adoption, debugging, noise, reporting limits, and ecosystem readiness remain major issues. |
| Lessons for builders | Platform PETs must be evaluated as product migrations: developer ergonomics and business utility can dominate cryptographic elegance. |
| Source | Privacy Sandbox Attribution Reporting overview and MDN Attribution Reporting API |
(Evidence: Deployment-backed. Source quality: Primary / official vendor documentation plus MDN platform documentation. Reviewed 2026-06-17 — browser support and ecosystem behavior are fast-moving; re-check before migration.)
Common Proposed Use Cases
| Use case | Candidate PETs | Why proposed | Caveats |
|---|---|---|---|
| Cross-platform campaign measurement | Clean rooms, attribution APIs, private aggregation | Advertisers need measurement after cookie and identifier loss | Noisy or thresholded outputs can hurt optimization |
| Audience overlap | PSI, clean rooms | Brands and platforms need match counts without exposing full lists | Match sets and repeated queries can leak |
| Incrementality testing | Clean rooms, DP, MPC | Parties need controlled experiment results | Experimental design can fail even if privacy controls work |
| Retail media collaboration | Clean rooms, DP, query governance | Retailers and brands need joint analytics | Platform incentives and output policy matter |
Lessons Learned
- Advertising PETs often trade user-level observability for aggregate measurement.
- Clean rooms do not automatically solve consent, purpose limitation, or platform power.
- Delays, noise, and thresholds are product constraints, not minor implementation details.
- Interoperability across platforms remains a practical barrier.