Deterministic vs. Probabilistic Match Rates: The Honest Numbers

Ask ten visitor-identification vendors for their match rate and you will get ten numbers that cannot be compared. Match rate is the headline metric of this category and the least standardized. This piece explains what the number actually measures, why deterministic and probabilistic rates are not the same thing, and how to pressure-test any vendor’s claim before you sign.

Deterministic vs. Probabilistic: The Core Distinction

A deterministic match is traced to verified, consented data — a known link between a cookie, a device ID, and a hashed email. It is either right or it is not resolved. A probabilistic match is a statistical inference: the system estimates that two signals probably belong to the same person or company, often from IP ranges, device characteristics, or behavioral patterns.

The difference is not academic. A probabilistic “match” can attach the wrong name to a session, and its accuracy degrades quietly as signals get noisier. Delivr.ai is 100% deterministic — zero probabilistic or modeled matching — and uses no device, browser, or canvas fingerprinting. Every resolved identity traces back to verified data.

Why Reported Rates Are Usually Inflated

Most inflated match rates come from three moves. First, quoting a company-level rate as if it were person-level — naming the logo is far easier than naming the buyer. Company-level rates run 30–50%; person-level rates run 10–20% in the US and under 5% in the EU for most tools. Second, quoting a rate against a favorable denominator — identifiable sessions rather than all traffic. Third, counting probabilistic guesses as matches, which lifts the number while lowering its reliability.

The result is a market where the number on the slide rarely survives a side-by-side test on your own traffic.

What a Real 50%+ Rate Means

Delivr.ai resolves 50%+ of anonymous visitors at the person level on US traffic, deterministically, from the first page view, with no gating. That rate comes from resolving cookies and mobile advertising IDs to hashed emails against an identity graph of 5.1B verified records and 491M US person profiles. It is a person-level number measured against real anonymous traffic, not a company-level number dressed up as one.

How to Verify Any Vendor’s Claim

• Ask what the rate measures: company or person? Only accept person-level as person-level.
• Ask for the denominator: all traffic, or a filtered subset?
• Ask deterministic or probabilistic — and what share of matches is each.
• Ask about geography: US and EU rates differ dramatically.
• Run a free match-rate test on your own traffic. The only number that matters is the one measured on your site, with your audience.