Somewhere in your company's strategy deck is a slide about first-party data. It is the asset, the moat, the answer to signal loss. What the slide never mentions is the number that decides whether any of it works: the match rate. Upload a customer file to an ad platform and only the records it can recognize become audience, suppression, or measurement. Everything else is a spreadsheet cosplaying as a strategy.
Industry-typical match rates run around twenty percent when files go in as-is. One in five. Sit with what that does to each downstream system. The lookalike model is learning your customer from a fifth of your customers, and not a random fifth; the records that match skew toward whoever gave you clean, current, personal email addresses, which biases the seed and therefore everything the model builds from it. The suppression list is failing to suppress four-fifths of your existing customers, so you are paying acquisition prices to advertise at people who already bought, then celebrating when some of them convert again. And the conversion matching behind your reported ROAS is built on the same sliver, which means the attribution debate your team has been having is a debate about the visible fifth of a mostly invisible business. I have watched companies argue attribution models for months while the match rate underneath them made the entire debate decorative.
Where the eighty percent goes
The gap has causes, and the causes have fixes, which is the good news hiding in a bad number. Data hygiene first: normalization, formatting, completeness, the difference between a CRM that collected emails and one that collected identities. The work email that bounced two jobs ago matches nothing. The phone number stored in three formats matches inconsistently. Collection design second, because a checkout that captures one weak identifier per customer has decided its future match rate at the point of sale; every additional identifier collected, phone alongside email, name alongside both, multiplies the chances a platform recognizes the human behind the record. Then the plumbing itself, how records are keyed, hashed, and delivered, where small technical choices swing results enormously: whether you normalize before hashing, which identifiers you send together, whether you use the platform's API or a file upload from 2019. This is exactly the territory where strategy without technical literacy gets owned, because the work looks like janitorial data labor and behaves like a media multiplier.
Here is the part that makes match rate unusual among marketing problems, and it is why I lead audits with it: fixing it improves every dollar you spend simultaneously, without touching a single campaign. Better creative helps the campaigns that run it. Better bidding helps the auctions it enters. A match rate that moves from twenty to sixty helps targeting, suppression, lookalikes, and measurement across every platform at once, permanently, for the cost of data work that most organizations already have the staff to do. There is no other lever in the account with that blast radius, and it is the least glamorous item on any roadmap it appears on, which is why it loses the prioritization meeting to whatever has a demo.
The ceiling moves
It is fixable at scale, and I say that from direct experience rather than vendor literature. In our own client work we have taken match rates from that industry-typical twenty percent to eighty, through exactly the unglamorous sequence above: hygiene, collection redesign, and plumbing done properly. The effect is not subtle. The same budget, the same creative, and suddenly four times the customer signal reaching every system that targets, suppresses, and measures. Reported performance improves partly because real performance improved and partly because measurement finally sees the business it was supposed to be measuring, and both improvements are real. Nothing else in the plan changed. The ceiling moved.
Before the next attribution argument, before the next audience workshop, before anyone says first-party data in a planning meeting again, ask one question. What is our match rate? If nobody in the room knows, that is the answer, and it is also the roadmap.