Website Visitor Identification Software: Compare Useful Signals

Run every website visitor identification vendor against the same week of your own traffic, then judge them on signals a rep actually acted on rather than on how many rows appeared in the dashboard. Reveal counts are the number vendor dashboards lead with, and a reveal nobody works produces no pipeline. The condition that decides the category is geography: person-level identification is largely a United States capability, so a team selling into the EU is choosing between account-level tools whatever the demo showed.

Fix the Traffic Set Before You Fix the Shortlist

Vendors are usually trialed sequentially, one month each. That design guarantees an unusable comparison, because your traffic in March is not your traffic in May, and a campaign launch or a conference mention will move coverage more than any difference between tools.

Run the candidates in parallel on the same pages over the same window. Two to four weeks is normally enough to accumulate a comparable set. Install every pixel at once, confirm each one fires on the same page inventory, and freeze the page set for the duration.

Before counting a single reveal, remove the traffic that should never have been in the denominator:

Google Analytics handles part of this automatically. Google documents that traffic from known bots and spiders is automatically excluded using a combination of Google research and the International Spiders and Bots List maintained by the Interactive Advertising Bureau, and that you cannot disable this exclusion or see how much was removed. Your identification vendor is a separate pipeline and makes no such promise, so ask each one directly what bot filtering it applies and whether filtered traffic is still billed.

Decide Account Identity or Person Identity First

This is the fork that determines your shortlist, and it is a workflow question, not a preference.

Account-level tools tell you an organization visited. That is enough to trigger account research, an account-based ad audience, or a note to whoever owns the account. Leadfeeder's pricing shows the model clearly: it identifies companies instead of individuals, and a unique company identified within the month counts once no matter how often it returns. Contact enrichment sits on higher plans and consumes credits separately.

Person-level tools name an individual. RB2B's pricing publishes basic resolution at 15 to 20 percent coverage globally and company-level only, with contact-level site identification available for US markets on Pro and Pro+, where premium resolution is published at 35 to 45 percent coverage.

The geographic constraint is the part to settle before demos. If most of your addressable market sits outside the United States, person-level pricing buys you account-level results, and you should be comparing account-level tools on account-level merits.

Our view: person-level identification is worth paying for only when a named individual changes the next action. If your play is an account-based ad audience or a research task, the extra spend buys precision you will not use.

Count Useful Signals, Not Reveals

A reveal is a row. A useful signal is a row that a human opened and acted on. The gap between those two numbers is what your cost per signal is actually measuring.

Define usefulness before the trial so the result cannot be rationalized afterwards. A workable definition: the identified account is in your ICP, is not an existing customer or open opportunity, visited a page that implies commercial interest, and a rep took a recorded action within the response window you set.

Then compute the number that matters:

cost per useful signal = monthly platform cost / signals meeting the definition

Run it per vendor on the shared traffic set. A tool producing four times the reveals at twice the price can still lose, because the additional volume tends to concentrate in traffic that fails the ICP filter.

Track false matches in the same pass. Sample fifty identified accounts per vendor and verify them by hand against the page and timestamp. False matches are rarely reported by vendors and they are expensive in a specific way: a rep researches and reaches out to a company that never visited, and the reputational cost lands on the signal source rather than the vendor.

Test Routing and Suppression, Not Just the Feed

A signal that arrives somewhere nobody looks is not a signal. During the trial, wire each vendor to the destination you would really use and measure what happens to the record afterwards.

Check that routing respects account ownership so a named account reaches its owner instead of a shared channel. Confirm that suppression lists actually suppress: load your customer list, your open opportunities, your competitors, and your own domains, then verify those accounts stop appearing instead of appearing with a label.

Verify the CRM write path before the contract, not during onboarding. Ask what object gets created, what happens when the account already exists, whether repeat visits append to a single record or create duplicates, and whether the tool can update without overwriting fields a human owns. Duplicate creation is the failure worth testing for, because it accumulates silently and is far harder to unwind than to prevent.

Price the Real Contract Shape

Published prices are entry points. Build the comparison from the terms that determine the invoice.

RB2B lists Free at $0 for 150 monthly resolutions, Starter at $79 for 300, Pro at $149, and Pro+ at $199, with overage at $0.45 per resolution on Starter and $0.25 on Pro and Pro+, plus additional domains at $99 per month. Leadfeeder publishes Lite at €0, Discover at €79 per month, Activate at €369, and Scale at €599 with annual billing only, tiered by the volume of companies identified.

Those structures behave differently as you grow. A per-resolution overage rewards tight page targeting, while a company-volume tier rewards nothing until you cross a threshold and then jumps. Model your expected and high traffic cases against both before choosing, and confirm what happens when you exceed the allowance mid-term: whether signals stop, queue, bill at overage, or force a plan upgrade.

Ask about export rights explicitly. Confirm you can export identified accounts with timestamps, page paths, and the identity evidence, and confirm what you may retain after termination. A tool whose history you cannot take with you makes every attribution question unanswerable the moment you switch.

LeadHaste practice: we keep the raw signal feed in a store we control, separate from the vendor and separate from the CRM, so a vendor change does not erase the history that justified the spend. That is our operating rule, not a feature any of these vendors provide.

Run the Decision in Sequence

1. Freeze the page set and install every candidate pixel together
2. Apply the exclusion list before counting anything
3. Settle account-level versus person-level on geography and workflow
4. Hand-verify a sample of matches per vendor
5. Compute cost per useful signal on the shared traffic set
6. Test routing, suppression, and CRM writes end to end
7. Price the expected and high cases against the real overage terms

The winner is usually not the tool with the highest reveal count, and the evaluation is only credible because every vendor saw the same visitors on the same days. If you are weighing a specific pairing rather than the category, our RB2B vs Warmly comparison works the same evidence from the other direction.

We can instrument the shared traffic set, define the usefulness test with your team, and run the vendor comparison during a free ICP and campaign-fit discovery call. Book your free ICP and campaign-fit discovery call →