The co-founders of a new privacy tech startup called Precise.ai are two names you’ll probably recognize: Adam Helfgott and Jesse Redniss.
Helfgott is a founder and the former CEO of TV ad platform Madhive, and Redniss was a data strategy executive at WarnerMedia and Turner before launching Qonsent in 2021, a solutions provider for user consent and privacy management.
On Wednesday, the duo announced the formation of Precise.ai, which merges Qonsent with blockchain-based data infrastructure technology developed by Valence Labs, an R&D hub for cryptography tech founded by Helfgott in 2021.
Redniss will be CEO of Precise, and Helfgott will serve as a strategic advisor on tech and engineering.
Doing the end-to-end thing
The big picture idea, Redniss told AdExchanger, is to combine privacy-compliant consent management with (deep breath) federated machine learning, custom audience creation using first-party data and contextual signals, identity verification, the ability to track what data is used where, measurement and activation across channels.
That’s a long list of some of ad tech’s favorite buzzwords, but it boils down to this: getting permission to collect data, keeping track of where it came from and allowing multiple parties to share, learn from and use it without having to move it anywhere, not unlike in a clean room.
If that sounds to you like Precise is looking to drink a lot of other people’s milkshakes, well, you’re correct.
Although the plan isn’t to disintermediate data clean rooms, customer data platforms and consent management platforms, the fact is “a lot of things in the marketplace are getting commoditized,” Redniss said.
“Many of those functions are being pulled into larger cloud operations, including Amazon, Snowflake and Databricks,” he said. “We’re trying to solve for many of these components in an end-to-end infrastructure, which is also the direction that the broader industry is moving in.”
The case for data decentralization
So, how exactly does Precise work in practice?
Qonsent brings consumer-facing tools that people can use to manage their permissions on a brand-by-brand basis.
Valence, meanwhile, has built data infrastructure on the blockchain that can verify data ownership and provenance, use encryption to track how data is being used and attribute the value of data back to specific sources.
Rather than a brand storing its data in a centralized database and working with multiple different partners to analyze and activate it, Precise would allow brands to onboard their data and make it accessible via “contributory nodes” within a network.
Through these connection points, brands could mix multiple data sets without moving or compromising them and query data across the network to get insights.
For example, if a fashion brand observes that certain customers are suddenly spending $200 more a week on clothing, other brands could use that information to deduce that these customers may have moved into a new economic bracket and could be in the market for high-ticket items, such as a new car or top-of-the-line sneakers.
Modeling real spending patterns derived from consented first-party data can provide a better-fidelity signal of intent than it’s possible to ascertain from, say, someone reading an article about buying a car, Helfgott said.
Creating a decentralized marketplace for exchanging data makes sense, he said, because it gives transparency to all the parties involved in a transaction without exposing PII.
But don’t call it “the blockchain.” That term has become reductive and borderline meaningless, Helfgott said, because too many companies bought into the hype and used “blockchain for blockchain’s sake, which created a lot of noise.”
“When Google Cloud comes out with a new product in their suite,” Helfgott said, “we’re not all out there saying, ‘You’re a BigQuery company’; people just use the product because it’s the best tool for the job from an engineering point of view. And that’s all that blockchain is: an engineering utility.”
For starters, Precise will attempt to slip as smoothly as possible into the existing ad buying process.
It will use first-party data to generate already available identifiers, including Ramp IDs and UID2s, and also integrate with a broad spectrum of SSPs, DSPs and ad servers.
“We’re being careful right now to support interoperability for the current way things are done,” Redniss said. “We don’t want to disrupt workflows on the execution side.”
But Precise is developing its own “more permissioned ID token structure,” he said, using the consent mechanisms developed for the historical Qonsent business. And it also has its own DSP for activation.
Precise will introduce these additional elements once more customers are up and running with the new data infrastructure. (Redniss said Precise is in the midst of signing up a large sports publisher, but declined to name it.)
As for current Qonsent clients, nothing changes for the moment. They can continue using the same tools to collect consented first-party data and do ID validation as usual, although Precise will try to upsell them on the new capabilities.
Redniss and Helfgott are still deciding whether they’ll keep or retire the Qonsent brand name going forward.