Results From a First-Party ID Pilot, Six Months In

A retail client wanted a first-party ID pilot because their paid media team was panicking about match rates on Meta and Google audiences. The pitch was simple: collect hashed emails and phone numbers at every touchpoint, pass them through a resolution layer, and hand vendors a cleaner identifier than a third-party cookie that's dying anyway.

Six months in, the match-rate improvement was real but modest: audience match rates on Meta went from around 58% to 71%. Nice, not transformative. The thing that actually moved the needle was internal — we finally had a single customer key that tied loyalty, support, and purchase data together without three separate join tables that disagreed with each other.

What the pilot actually measured

• Audience match rate on Meta and Google, before and after, measured monthly to control for seasonal noise.
• Email hash coverage rate across the funnel — what percentage of sessions had a resolvable identifier by checkout.
• Internal join accuracy between CRM and purchase records, measured by manual spot-check against a 500-record sample.
• Resolution latency — how long it took a new hashed identifier to propagate to downstream systems.

We scoped the pilot to fix an ad-targeting problem and it ended up fixing a data-warehouse problem nobody had officially complained about.

The part nobody warns you about

Resolution latency was the surprise pain point. A new hashed email didn't propagate to the ad platforms for up to 48 hours in our setup, which meant retargeting a user who'd just abandoned a cart during the highest-intent window simply didn't work for two days. We fixed it by prioritizing a faster sync path for checkout-abandonment events specifically, rather than trying to make the whole pipeline faster — a targeted fix instead of an expensive general one.

If you're evaluating a first-party ID pilot, measure the internal data-quality benefit as seriously as the ad-targeting benefit. In our case the internal benefit is what got budget approved for phase two; the match-rate number was just the thing that got the meeting scheduled.