Today, all companies and regulators talk about Privacy Enhancing Technologies (PETs), that is to say popular privacy tools such as differential privacy, federated learning, homomorphic encryption or Secure Multiparty Computation (SMPC). Those are brilliant tools to frame data collection. They help companies hide or scramble data so no one can snoop on it.

But here’s the problem nobody talks about: in real‑time ad auctions (RTB), you have less than 100 milliseconds to decide if an ad should be shown. That’s faster than a human blink. These privacy tools are simply too slow for that speed.
- Homomorphic encryption? Up to 10,000 times slower.
- Federated learning? Can leak user patterns anyway.
- Differential privacy? Works for “what do a million people like?”, not for “does this one person say yes to this ad right now?”
So companies end up with great privacy assets on paper… that fail in real life when the bid request comes in.

The question to address is not HOW to “hide the data forever.” Instead, it is: "What uses are authorised by the user, per request?". That’s request‑based de‑identification. This is what actually works in Real‑Time AdTech, because it:
- Uses HMAC (a simple, fast, proven maths function),
- Runs easily inside 100ms,
- Works with existing ad infrastructure (no complete rebuild),
- Gives users individual control over each type of data use.

ID side State‑of‑the‑Art Solution doesn’t sacrifice real‑time performance to achieve privacy and consent. While others focus on protecting data at rest, we protect each decision in flight (at auction speed, at scale, with real user consent built into every request). That’s not just better privacy. That’s privacy that actually works in the real world of AdTech.

So, anticipate article 88B GDPR and start exploring how this works with your current ad stack!