Connecting ad exposure to a downstream conversion through a deterministic or modeled path, proving the ad caused the outcome.
Deterministic identity matching across UID2, RampID, and publisher-provided signals still outperforms probabilistic methods by 3–5x on CTV inventory.
Clean-room integration enables closed-loop attribution within a white-label DSP without exposing raw user-level data to either party.
| Criterion | What it separates | Evidence |
|---|---|---|
| Identity graph coverage | Deterministic vs. probabilistic match rate across channels | doc · tested |
| Latency to attribution | Hours or days from conversion event to attributed report | doc · tested |
| Incrementality methodology | Ghost bids, PSA holdouts, geo-experiments, or self-reported lift | doc · doc |
| Privacy compliance posture | Clean-room vs. raw-log join, and which regulations the method survives | contract · doc |
| Cross-channel support | Display, video, CTV, audio, social — which paths are covered | integration list · doc |
| Output portability | Can the attributed data leave the platform and feed another system | doc · doc |