Direct answer: Across 1,232,401 real B2B website sessions measured over 90 days on 11 sites, company-identified sessions submitted a form at 3.4x the rate of anonymous sessions, linked to a CRM contact at 4.5x the rate, and returned to the site at 3.7x the rate. Strip out the sessions that were only "identified" because the company was already a known CRM contact, and the honest, IP-identification-only form rate gap is 2.4x, not 3.4x. Identification does not cause that gap. It concentrates a population that was always more likely to convert, and correctly used, that concentration is a prioritization signal worth building a motion around. See it live to check where your own traffic falls on this curve.
Key takeaways
- 562,937 of 1,232,401 sessions (45.7%) were company-identified; 669,464 (54.3%) stayed anonymous.
- Identified sessions submit a form at 3.67% versus 1.07% for anonymous sessions, a 3.4x gap. Excluding sessions already known from the customer's CRM, the honest IP-identification-only gap is 2.52% versus 1.07%, or 2.4x.
- Identified sessions link to a CRM contact at 10.11% versus 2.23% for anonymous (4.5x), and return to the site at 50.5% versus 13.8% (3.7x).
- Within identified sessions, match confidence matters more than the identified/anonymous split itself: the highest-confidence tier converts at 7.55%, roughly 7x the anonymous rate, while the lowest-confidence tier converts at 1.46%, barely above anonymous.
- This is a correlation, not a causal lift. Anonymous traffic includes ISPs, VPNs, datacenter proxies, and bots that were never going to convert regardless of whether they were identified.
How we measured it
This is our second data study built from first-party production traffic on the Abmatic AI platform, following our visitor identification match rate study. That first study asked how often a session resolves to a company or person. This one asks the next question: once a session is identified, does it actually behave differently, and by how much?
We looked at 1,232,401 real, non-simulated sessions across 11 B2B websites running Abmatic AI, over a rolling 90-day window ending July 14, 2026. A session counts as company-identified if it had a non-empty company match, by name or domain. Form submission and returning-visitor status are visitor-level outcomes attached to the session's visitor, not the single session. No customer, site, or identification provider is named anywhere in this study. All figures are aggregated percentages and ratios.
Finding 1: Identified sessions submit forms at multiples of the anonymous rate, but the honest multiple is smaller than the flattering one
The headline comparison: identified sessions submitted a form at 3.67%, anonymous sessions at 1.07%, a 3.4x gap. That is the number a dashboard shows you by default, and it overstates the effect of identification itself. A subset of "identified" sessions, 14,800 of them, or 2.6% of all identified sessions, were companies the customer already knew from their own CRM rather than companies identification newly surfaced. That subgroup behaves like existing pipeline: a 46.1% form rate and 90.7% returning-visitor rate, which pulls the blended average up.
Exclude that CRM-known subset and look only at sessions identification actually surfaced from IP, and the form rate gap is 2.52% versus 1.07%, a 2.4x multiple. Present 2.4x as the honest, IP-identification-only number. The 3.4x figure is real but only holds with the CRM-known caveat attached; without that caveat it silently credits identification for conversions that were never identification's doing.
| Cohort | Form submission rate | Multiple vs anonymous |
|---|---|---|
| Anonymous sessions | 1.07% | 1.0x (baseline) |
| IP-identified sessions (CRM-known excluded) | 2.52% | 2.4x |
| All identified sessions (CRM-known included) | 3.67% | 3.4x |
| CRM-known subset only | 46.1% | 43.1x |
Finding 2: CRM linkage and return visits show the same pattern, a real gap, inflated by known accounts
CRM-linked contact rate was 10.11% for identified sessions versus 2.23% for anonymous, a 4.5x gap. Returning-visitor rate was 50.5% for identified versus 13.8% for anonymous, a 3.7x gap. Both numbers are blended across the full identified population, including the CRM-known subset described above, so the same honesty adjustment applies: a meaningful share of the CRM-linkage and return-visit gap is explained by companies that were already known, returning customers, or existing pipeline being present in the "identified" bucket by definition. Treat 4.5x and 3.7x as the ceiling, not the typical case for a newly IP-identified anonymous visitor.
Finding 3: Page depth barely moves; the gap is in what happens after the visit, not during it
If identification were simply flagging more engaged browsing sessions, you would expect a large gap in on-site behavior. It is small. Average page views per session were 1.06 for identified versus 0.87 for anonymous. Multi-page sessions were 2.4% of identified versus 1.1% of anonymous, a 2.2x gap, and deep sessions of five or more pages were 0.4% versus 0.1%. Identified visitors browse only slightly more than anonymous ones. The real gap opens after the session, in whether that visitor ever submits a form, gets linked to a CRM contact, or comes back, not in how many pages they click through on a given visit.
| Metric | Identified | Anonymous | Multiple |
|---|---|---|---|
| Avg page views per session | 1.06 | 0.87 | - |
| Multi-page sessions | 2.4% | 1.1% | 2.2x |
| Deep sessions (5+ pages) | 0.4% | 0.1% | - |
| Returning-visitor rate | 50.5% | 13.8% | 3.7x |
Skip the manual work
Abmatic AI runs targets, sequences, ads, meetings, and attribution autonomously. One platform replaces 9 tools.
See the demo →Finding 4: Confidence tier predicts conversion better than identified-or-not does
Splitting IP-identified sessions by match confidence tier tells the sharper story. The lowest confidence tier, 220,235 sessions, converted at 1.46%, barely above the anonymous 1.07% baseline. The highest confidence tier, 77,544 sessions, converted at 7.55%, roughly 7x the anonymous rate and 4-5x the low-confidence tiers. The tiers in between, at confidence scores of 30, 50, and 70, converted between 1.83% and 1.96%, clustered well below the top tier and only modestly above the bottom.
| Confidence tier | Sessions | Form rate | Returning |
|---|---|---|---|
| 20 (lowest) | 220,235 | 1.46% | 48.9% |
| 30 | 115,406 | 1.96% | 50.1% |
| 50 | 98,950 | 1.85% | 50.9% |
| 70 | 36,002 | 1.83% | 45.4% |
| 90 (highest) | 77,544 | 7.55% | 49.7% |
The practical read: the identified-versus-anonymous split is the wrong lens for prioritization on its own. Confidence tier is the sharper lens. Our first study found that only 20.9% of company matches are high confidence. That 1-in-5 slice is where the 7.55% conversion signal actually lives; the other four-fifths of "identified" sessions convert closer to anonymous than to that top tier. If a team routes every identified account to sales with no confidence filter, most of what gets routed will not outperform an unidentified visitor by much.
Why this is not a causal claim
None of the numbers above prove that identification causes a visitor to convert. Identification is a measurement, not an intervention: it observes who a visitor's company already is, it does not change the visitor's intent. The gap exists for a simpler reason. Anonymous traffic is not a clean sample of "the same buyers, just unlabeled." It includes ISPs, consumer VPN exit nodes, datacenter proxies, and bot traffic that will never fill out a form or return, no matter what identification technology is applied to it. Identified traffic, by construction, skews toward real companies on real corporate networks, the population that was always more likely to have a genuine buyer behind it.
The correct frame is prioritization, not lift. Identification does not manufacture demand; it tells you where the demand that already exists is concentrated, so a sales or marketing team can spend attention on the accounts most likely to already be evaluating, instead of spreading effort evenly across a traffic mix that is, in large part, never going to convert.
What this means if you are building a motion on identified traffic
Three practical adjustments follow directly from the data. First, report the honest number: use 2.4x for IP-identified sessions when talking about the lift identification itself produces, and disclose the CRM-known subset separately rather than blending it into a single headline multiple. Second, filter by confidence tier before routing anything to a rep; the data above shows the top tier converting 4-5x higher than the tiers just below it, so a raw "identified" flag without a confidence threshold wastes attention on a population barely different from anonymous. Third, do not expect identification alone to move browsing depth much, page views per session move only slightly; the return on identification shows up downstream, in whether a visitor eventually submits a form or reappears, which argues for pairing identification with a nurture or personalization motion rather than treating the identification event itself as the win.
This is the same conclusion our reverse IP lookup coverage reaches from the technical side: a match is an input to a workflow, not an outcome. Abmatic AI pairs visitor identification with confidence-tier routing and on-site personalization in one platform, starting at $36K/year, so the highest-confidence accounts trigger action instead of sitting in a spreadsheet. Book a demo to see how your own traffic splits by confidence tier before you build a routing rule around it.
Frequently Asked Questions
Do identified website visitors actually convert more than anonymous ones?
Yes, but the honest multiple is smaller than the headline one. Across 1,232,401 sessions, identified visitors submitted forms at 3.67% versus 1.07% for anonymous, a 3.4x gap. Excluding sessions that were only identified because the company was already a known CRM contact, the IP-identification-only gap is 2.52% versus 1.07%, a 2.4x multiple. Use 2.4x when describing what identification itself contributes.
Does identifying a visitor cause them to convert?
No. Identification is a measurement of who a visitor's company already is, not an intervention that changes buyer intent. The observed gap is largely selection bias: anonymous traffic includes ISPs, VPNs, datacenter proxies, and bots that were never going to convert, while identified traffic skews toward real companies on corporate networks. Treat identification as a prioritization signal, not a causal lift.
Does match confidence matter more than whether a visitor is identified at all?
In our data, yes. The highest-confidence tier (90) converted at 7.55%, roughly 7x the anonymous rate, while the lowest-confidence tier (20) converted at 1.46%, barely above anonymous. Routing every "identified" session to sales without a confidence filter mostly routes sessions that convert close to the anonymous baseline.
How many B2B website sessions actually get identified?
In this dataset, 45.7% of sessions (562,937 of 1,232,401) were company-identified and 54.3% stayed anonymous. See our separate visitor identification match rate study for company- and domain-level match rates across a different B2B session sample.
What should I do with this data if I am prioritizing which identified accounts to act on?
Filter by match confidence before routing to sales or triggering outbound. The highest-confidence tier in our data converted 4-5x higher than the tiers directly below it. See our account match rate explainer for how confidence scoring works, and pair identification with on-site personalization so the highest-confidence accounts get acted on rather than just logged.
The honest summary
Company-identified B2B sessions submit forms, link to CRM contacts, and return to the site at real multiples of the anonymous rate, and the honest version of that multiple, with known accounts excluded, is 2.4x on form submissions. But the gap is not proof that identification changes behavior; it is proof that identification finds a population that was already more likely to convert, buried inside a traffic mix that includes a large share of visitors who never will. The number worth building a motion around is not "identified versus anonymous." It is match confidence, where the top tier in our data outconverts anonymous traffic by roughly 7x. Measure that, filter on it, and identification stops being a vanity metric and starts being a prioritization engine. Book a demo to see your own confidence-tier breakdown.
