A realistic company-level match rate for website visitor identification sits somewhere around 30 to 65 percent of qualifying B2B traffic, and person-level match rates run much lower, often in the low double digits in the US and lower still in the EU. Any vendor quoting you a flat "we identify 70 percent of your visitors" is almost certainly measuring against a denominator that flatters them. The number itself is meaningless until you know what it is a percentage of.
This guide explains what match rate actually measures, the three denominators vendors quietly swap between, the ranges you should expect for company-level versus person-level identification across the US and EU, what drives the number up or down, and how to run an apples-to-apples test so two vendors can be compared honestly. The hard part is not the math. It is that no two vendors define the term the same way, so headline comparisons are usually fiction.
Book a demo to see how Abmatic AI reports match rate on your real traffic, with the denominator shown, so you can evaluate it on the same rigorous test you give every other tool.
What "Match Rate" Actually Means
Match rate is a fraction. The numerator is the number of visits (or visitors, or accounts) a tool resolved to an identity. The denominator is the traffic you are dividing by. Vendors agree on roughly what the numerator is. They almost never agree on the denominator, and that is where the inflation lives.
Three denominators get used interchangeably, and they produce wildly different numbers from the same underlying performance:
- Percent of all traffic. Every visit, including bots, your own employees, returning logged-in customers, and consumer ISP traffic. This is the smallest, most honest-sounding denominator and the hardest to hit a high number against.
- Percent of company traffic. Only the visits a tool already believes are business visitors. This quietly removes residential and mobile traffic from the denominator before the division happens, which inflates the result.
- Percent of eligible or resolvable traffic. The narrowest denominator. The vendor defines "eligible" (often: US-based, business network, not a bot, consented) and reports the match rate of that slice. A 90 percent match rate against a denominator you cannot see is a marketing number.
The same tool can truthfully claim 22 percent, 48 percent, and 88 percent for the same week of traffic depending on which denominator it picks. None of those numbers is a lie. Only one of them tells you how much of your actual pipeline the tool will surface. When you read a match-rate claim, your first question is always the same: a percentage of what?
Company-Level vs. Person-Level Match Rate
These are two different products that share a name. Company-level (account-level) identification resolves the organization behind a visit, usually through reverse IP lookup against an IP-to-company database. Person-level (contact-level) identification names the individual human, which requires identity-graph or first-party signals on top of the IP match. The difference matters because a vendor will often quote the company-level rate, which is much higher, and let you assume it applies to contacts. It does not. The split between these two is covered in depth in our guide to contact-level vs. account-level de-anonymization.
Realistic Match-Rate Ranges by Type and Geo
The ranges below are typical, illustrative bands drawn from how these methods perform in practice, not a citation from a specific study. Treat them as a sanity check: if a vendor's claim sits far above the top of a band, ask what they changed about the denominator. All figures assume B2B traffic. Consumer-heavy sites behave differently and usually match worse at the company level.
| Identification type | Region | Typical match rate (% of B2B traffic) | What inflates the headline |
|---|---|---|---|
| Company-level (reverse IP) | US | ~40 to 65% | Quoting % of "company" traffic only |
| Company-level (reverse IP) | EU | ~30 to 55% | Smaller dedicated-range coverage, consent gating |
| Person-level (identity graph) | US | ~10 to 30% | Quoting against US-only, consented traffic |
| Person-level (identity graph) | EU | ~2 to 12% | GDPR limits graph coverage sharply |
| First-party (returning known visitors) | US / EU | ~15 to 40% of returning traffic | Counting only previously identified visitors |
Two things jump out. First, person-level rates are a fraction of company-level rates, so conflating them overstates contact coverage by a wide margin. Second, the EU gap is real and structural, not a tuning problem. Identity graphs in Europe are thinner because consent rules restrict the data that feeds them, so a tool that hits 25 percent person-level in the US can fall to single digits in Germany or France. If most of your traffic is European, weight your evaluation accordingly and read whether website visitor de-anonymization is GDPR compliant before you build a motion on it.
What Actually Drives Your Match Rate
Two companies running the same tool will get different match rates because match rate is as much about your traffic as it is about the vendor. Before you blame or credit a tool, account for these.
Traffic Mix
The single biggest factor. A site whose visitors are mostly remote workers on home broadband will match far lower than one whose visitors sit in corporate offices on registered IP ranges. Mobile traffic, where carrier-grade NAT shares one IP across thousands of users, is nearly impossible to attribute to a company. If half your traffic is mobile, half your traffic is effectively invisible to IP-based matching no matter how good the vendor is.
B2B vs. B2C Skew
IP-to-company databases are built around business networks. The more your audience looks like consumers on residential ISPs, the lower the company-level match. A pure B2B SaaS site can clear the top of the ranges above. A site that mixes in heavy consumer or SMB-on-home-internet traffic will not, and that is a property of the audience, not a vendor failure.
Bot and Internal-Traffic Filtering
This cuts both ways. Aggressive bot filtering lowers the headline match rate because bots are easy to "identify" as data centers, and removing them shrinks an easy-win bucket. But not filtering bots and internal traffic produces a junk match rate full of crawlers and your own employees. When you compare vendors, make sure they are filtering the same things, or one will look better purely because it counts noise.
Identity-Graph Freshness and Reach
For person-level matching, the vendor's identity graph is the engine. A larger, more frequently refreshed graph resolves more individuals. A stale or small graph misses them. This is also the least transparent input, because no vendor will show you the inside of their graph. You judge it by output on your traffic, not by their description of it.
Consent and Region
Consent banners that gate tracking until acceptance remove a slice of traffic from the resolvable pool. In the EU this is large. A tool can be performing well on the traffic it is allowed to see and still report a low rate against all traffic, because a big chunk was never eligible.
How Vendors Quote Match Rate (and the Games They Play)
Once you know the denominators, the patterns become easy to spot. Here is how the same real performance gets dressed up.
| How it's quoted | What they did | What to ask |
|---|---|---|
| "We identify 70%+ of visitors" | Used % of company traffic, dropped residential/mobile from the denominator | Is that 70% of all traffic or of pre-filtered business traffic? |
| "Industry-leading match rate" | No denominator stated at all | Show me the formula. Numerator over what? |
| "Up to 90% accuracy" | Mixed up accuracy (is the match correct?) with match rate (how much got matched?) | Are you quoting coverage or correctness? They are different metrics. |
| "We match more people than competitor X" | Counted company-level matches as "people" | Is this unique persons, or accounts you're calling people? |
| "60% match on US traffic" | Picked the region where they perform best | What's the rate on my actual geo split? |
Accuracy versus match rate is the trap that catches the most buyers. Match rate is coverage: what share did the tool resolve? Accuracy is correctness: of the ones it resolved, how many are right? A tool can have a high match rate and poor accuracy (it labels lots of visits, many wrongly) or a low match rate and high accuracy (it only commits when sure). You need both numbers, and a vendor that offers one and stays quiet on the other is choosing the flattering one. For a broader read on how the major tools stack up on these claims, see our comprehensive review of visitor de-anonymization tools and the best website de-anonymization tools for 2026.
Skip the manual work
Abmatic AI runs targets, sequences, ads, meetings, and attribution autonomously. One platform replaces 9 tools.
See the demo →How to Run an Honest, Apples-to-Apples Vendor Test
The only match rate that matters is the one measured on your traffic, with a denominator you define, across every vendor at once. A vendor's own dashboard number is their measurement, not yours. Run the test yourself.
- Fix one denominator and force every vendor to use it. Decide up front: match rate equals identified visits divided by total qualifying B2B visits, where you define "qualifying" (exclude bots, internal IPs, known consumers). Hand every vendor the same definition. Refuse their preferred denominator.
- Run all vendors on the same traffic, same window. Install the candidate pixels in parallel for two to four weeks so they all see identical visits. Comparing one vendor's last quarter to another's pilot week is comparing nothing.
- Separate company-level from person-level in the results. Report two numbers per vendor. Never let a high company rate stand in for the contact rate you actually want.
- Spot-check accuracy by hand. Pull a random sample of identified visits and verify the company or person is correct. A 50 percent match rate that is 95 percent correct beats a 70 percent rate that is 60 percent correct.
- Segment by geo and device. Break results into US versus EU and desktop versus mobile. This reveals whether a vendor's headline number came from cherry-picking your best-matching segment.
- Translate to cost per identified account. Match rate only matters relative to price. A higher rate at a higher price can lose to a lower rate that costs less per resolved account. Pair this test with our visitor identification pricing guide so you compare value, not just coverage.
If a vendor resists running on your traffic with your denominator and pushes you toward their case-study number, that resistance is the answer. A tool confident in its match rate will let you measure it yourself.
Red Flags in Vendor Match-Rate Claims
- A single headline percentage with no denominator. Always inflated until proven otherwise.
- Refusing a side-by-side pilot on your own traffic. They want you to trust their measurement instead of making your own.
- Quoting person-level coverage that rivals company-level. Structurally implausible; usually company matches relabeled as people.
- One global number with no geo breakdown. Hides a weak EU rate behind a strong US one.
- "Accuracy" and "match rate" used as if they are the same thing. Either confusion or sleight of hand.
- No mention of bot or internal-traffic filtering. The rate may be padded with crawlers and employees.
How Abmatic AI Reports Match Rate
Abmatic AI is an AI-native ABM and revenue platform that identifies both the companies and the individual contacts behind anonymous traffic, then lets you act on that identity in one place. On the specific question of match rate, the position is simple: we will measure it on your traffic, show you the denominator, and split company-level from contact-level so the number means something.
What that looks like in practice:
- Reverse-IP plus first-party identity resolution. Company-level matching covers the office traffic, and first-party signal capture across web, ads, LinkedIn, and email pushes resolution past the company boundary to the person, which is where the contact-level rate comes from.
- Honest denominators. Match rate is reported against the traffic you define, not a hidden "eligible" slice, so what you see in a pilot is what you get in production.
- Two numbers, not one. Account-level and contact-level coverage are reported separately, because they are different products and pretending otherwise would be exactly the game this guide warns about.
- From match to motion. A resolved identity is only useful if it triggers something. Identified accounts and contacts flow into web personalization, agentic outbound, agentic chat, ad orchestration, and your CRM through bi-directional Salesforce and HubSpot sync, so coverage turns into pipeline instead of a log file.
Abmatic AI is built for mid-market through enterprise B2B teams, with pricing starting at $36,000 per year and enterprise tiers available. Because it is first-party-first, you can stand up a real match-rate test in days. Give it the same denominator and the same traffic window you give every other vendor, and judge it on the result.
See it on your traffic: book a demo and run Abmatic AI on the same rigorous, apples-to-apples test this guide describes.
Frequently asked questions
What is a good match rate for website visitor identification?
For company-level identification on US B2B traffic, roughly 40 to 65 percent of qualifying visits is a realistic, healthy range. For person-level identification, expect far less, often 10 to 30 percent in the US and lower in the EU. There is no single "good" number, because it depends entirely on the denominator and on your traffic mix. A rate is only meaningful when you know what it is a percentage of.
Why do vendors quote such different match rates?
Because they use different denominators. The same performance can read as 22 percent of all traffic, 48 percent of business traffic, or 88 percent of "eligible" traffic. Vendors pick the denominator that produces the most impressive number. Always ask whether a quoted rate is a percentage of all traffic or of a pre-filtered slice.
How accurate is website visitor identification?
Accuracy and match rate are separate metrics. Match rate is how much traffic gets resolved; accuracy is how often the resolution is correct. Company-level matching is most accurate for corporate office networks on registered IP ranges and least accurate for remote, mobile, and VPN traffic. Always ask a vendor for both numbers, since a high match rate with poor accuracy is worse than the reverse.
Why is the EU match rate lower than the US?
Two reasons. Company-level IP coverage is somewhat thinner outside the US, and consent rules restrict the data that feeds person-level identity graphs. GDPR in particular limits the graph data available, so person-level match rates in the EU often fall to single digits. If your traffic is mostly European, weight your evaluation toward what the tool can resolve under those constraints.
How do I compare two visitor identification vendors fairly?
Run them in parallel on the same traffic for the same two-to-four-week window, hand both the same denominator definition, and report company-level and person-level rates separately. Then hand-verify a random sample for accuracy and convert each to cost per identified account. A vendor's own dashboard number is their measurement, not a fair comparison.
Can a tool identify the individual person behind a visit, not just the company?
Yes, but at a much lower match rate than company-level identification, and it requires identity-graph or first-party signals on top of reverse IP. Reverse IP alone returns the organization, not the human. Platforms like Abmatic AI combine reverse IP with first-party identity resolution to resolve contacts where consent and signal allow, and report that contact-level rate separately from the company-level rate.
