Search "RB2B match rate" and you'll find at least four different numbers: 5 to 20%, 15 to 45%, an unverified "40%+," and vendor-reported figures that never specify what they're a percentage of. None of these are lying exactly. They're measuring different things, on different traffic, against different denominators. The only match rate that means anything for your business is the one you calculate on your own site, and you can get it in about two weeks without switching tools.

How RB2B Actually Identifies a Visitor

RB2B runs on a JavaScript pixel that matches an anonymous session to a LinkedIn profile in real time. When someone lands on your site, the pixel captures the session and checks it against an identity graph built primarily from LinkedIn signals. If it finds a match, a Slack notification lands within minutes: name, title, company, LinkedIn URL, and the pages that person viewed.

Two structural facts about that mechanism explain almost everything else in this article. Person-level matching is US-only: due to GDPR and other international privacy constraints, RB2B's own documentation and independent reviews are consistent that person-level identification is limited to US IP addresses, with company-level identification at best for traffic from elsewhere. And matching depends on LinkedIn coverage: a session can only resolve to a profile if the visitor has an active LinkedIn account and enough signal for the match to fire, which works well for buyers in software and professional services and far worse in manufacturing, healthcare, trades, and other verticals where LinkedIn adoption is thinner.

Neither of these is a knock against RB2B specifically. Almost every identification approach has a dependency like this somewhere. It's exactly why match rate isn't a fixed property of a tool. It's a property of the tool applied to your traffic.

Why the Published Numbers Don't Reconcile

If you've read three "RB2B review" articles, you've read three different match rates. That's not sloppy research. It's the predictable result of three things review sites rarely disclose.

The denominator is almost never the same. Some numbers are calculated against every raw session that hits the site. Others are calculated only against sessions that already show buying intent, or only against US-based traffic to begin with. A 45% match rate against a narrow, pre-filtered slice of sessions and a 15% match rate against everything that touched the domain can both be technically true at the same time.

Traffic composition swings the number more than the tool does. A company selling to VP-of-Sales buyers at Series B SaaS startups will see a very different match rate than one selling to plant managers at industrial manufacturers, using the exact same tool. Review sites rarely disclose whose traffic they tested against.

And most of these numbers come from an interested party. A "best RB2B alternatives" listicle written by a competing vendor, or an affiliate site paid per lead, has little incentive to publish a rigorous, independently audited number. That doesn't make every figure fabricated. It does mean none of them should be treated as your number. We've written before about how to vet a vendor's accuracy claims in general terms. RB2B is simply the clearest current example of why that discipline matters: the spread between the lowest and highest published number for the same tool is nearly 10x.

Don't Trust the Number Until You've Run It On Your Own Traffic

We've published the full step-by-step process for this elsewhere, a concurrent two-tool test, engaged sessions as the denominator, matched against a spot-check of visitors you already know, in how to vet a website visitor ID vendor's accuracy claims. That process applies to RB2B exactly as written: run it alongside whatever else you're evaluating for a 14-day window, divide identified visitors by engaged sessions, and use that number instead of any of the four in the intro above.

What's specific to RB2B is knowing which two variables to watch while you run it: what share of your traffic is outside the US (person-level matching won't apply there at all), and what share of your ICP is active on LinkedIn (matching quality tracks LinkedIn usage closely, by design).

What This Test Usually Reveals

If your ICP traffic skews outside the US, or outside industries with heavy LinkedIn usage, expect your real number to land near the bottom of any published range, not the top a demo call quoted you. If most of your buyers are US-based software or professional services people who are active on LinkedIn, a LinkedIn-dependent tool can perform close to its best-case numbers.

This is also where the difference between person-level and company-level identification matters most. A tool that can't resolve a person can often still resolve the company, and for some plays that's enough. For others, especially anything routed to an individual rep for direct outreach, it isn't.

Knock2 approaches identification differently: instead of relying on a single platform's login state, it matches visitors against an identity graph built from a consent-based publisher network. Measured the exact same way as the test above (person-level identified visitors ÷ engaged sessions, same GA definition), Knock2 publishes 93%* company-level and 62%* person-level identification for US traffic. *Identification rates measured against engaged sessions. Results may vary by traffic profile, geography, and industry.

That's not an argument that Knock2 is automatically the right fit and RB2B isn't. It's an argument for running your own test before either number changes your budget. If you've already run the test and know you need a change, our head-to-head RB2B comparison lays out the switch in more detail.

What to Weigh Once You Have Your Number

Once you know your actual match rate, here's what should carry the most weight in deciding whether to switch, and how much:

• 🔴 Very High - Denominator transparency: does the vendor disclose exactly what their published rate is measured against.
• 🔴 Very High - Single-source dependency: is identification reliant on one platform's cookie or login state, or resolved against multiple sources.
• 🟠 High - Vertical fit: does your ICP sit in industries with strong LinkedIn adoption, or ones where it's thin.
• 🟡 Medium - Geographic coverage: US-only person-level identification versus broader coverage, if your buyers aren't all domestic.
• 🟢 Low-Medium - Workflow depth: does identification flow straight into your GTM stack and CRM, or stop at a Slack alert.
• ⚪ Low - Price per credit: matters far less once you actually know your real match rate.

FAQ

Is RB2B accurate outside the United States?

Not at the person level, by design. Person-level identification is restricted to US IP addresses because of GDPR and other international privacy constraints. International traffic gets company-level identification at best.

Can I run two website visitor identification tools at the same time?

Yes. Most identification pixels are independent JavaScript snippets that don't conflict. Running two in parallel for a two-week window is the fastest way to get a real side-by-side match rate without committing to a switch.

How does Knock2 measure its own match rate?

Knock2 publishes 93%* company-level and 62%* person-level identification for US traffic, both measured against engaged sessions (any visit lasting 10 seconds or longer, or including 2 or more pageviews, per Google Analytics' definition). *Results may vary by traffic profile, geography, and industry.

Match rate isn't a fact about a vendor. It's a fact about a vendor applied to your traffic. Every published number you'll find for RB2B, or for any identification tool, was measured on someone else's visitors, with someone else's denominator, often by someone with a reason to round in one direction. The two-week test above costs nothing but a little patience, and it's the only version of the number that should ever influence a renewal or a switch decision.

If your test shows you're missing a meaningful share of person-level matches, especially outside the US or outside LinkedIn-heavy verticals, it's worth seeing how an identity-graph approach compares on your own traffic. Start a free trial and run it yourself.