Every identity-based media plan begins with subtraction. Before an impression serves, the audience passes through an onboarding match, an identity graph, and a bid-stream lookup — and every one of those hops keeps only a fraction of what entered it. The industry reports each hop's rate separately, which hides the only number that matters: what is left at the end of the chain.
How match loss compounds
Illustrative — representative rates chosen to show the shape of the math; substitute your own vendor's numbers:
| Hop | Representative rate | Cumulative audience reached |
|---|---|---|
| Onboarding match | 60% | 60% |
| Identity-graph coverage | 70% | 42% |
| ID present in bid stream | 50% | 21% |
Twenty-one percent. The chain is multiplicative, so improving any single hop moves the outcome less than its vendor's deck implies — and every hop trends the wrong direction as identifiers deprecate.
The effective-CPM math
Same illustrative rates: a $6.00 data CPM spread over the 21% of the audience actually reached is an effective $28.57 per thousand genuinely in-audience impressions. The line item said six dollars. The plan paid twenty-eight.
What "no match required" actually means
A geographic segment has no chain, because the segment is the geography. Census block groups that qualify for the cohort are addressable in full — the reach figure at activation is the reach figure in market, and it does not decay between quarters, because nothing in it depends on an identifier surviving.
The precision objection, answered
Precision in a geographic system is a dial, not a casualty. The threshold index sets how strongly an area must over-index before it qualifies — raise it and the audience sharpens, lower it and it scales — and the tradeoff is visible before a dollar moves, not discovered in a post-campaign match report.
When ID-based still wins
Honestly: three places. Suppressing known customers, frequency-capping logged-in inventory, and closed-loop measurement against a CRM. Keep identity for the jobs where you genuinely know the individual — and stop paying the chain tax on prospecting reach, where you never did.
Common questions
What is a good match rate in 2026?
It is the wrong question. One hop's rate says nothing about delivered reach — the product of every hop does. Ask vendors for effective reach after onboarding, graph and bid-stream loss.
Is ID-free targeting less precise?
It is differently precise: the unit is the qualifying geography rather than the claimed individual, and the threshold index tunes how sharp that qualification is — before spend, not after.
Should I run ID-based and geographic together?
Most buyers do: identity for suppression, capping and measurement on known customers; geographic cohorts for prospecting reach that no longer depends on a match.