First-party data in B2B is what a person or a company hands you directly: a demo form, a gated download, a login, a CRM record your sales team created. The question first party data b2b teams actually face is not whether their data is “better” than third-party data — it is whether they can prove, for any specific person shown to sales, where that identity came from. If you cannot produce that trail, you have a guess with a nice interface.
Website visitor identification splits into two problems. Naming the company behind a visit matters for routing and account scoring. Naming a person inside that company touches an identifiable individual.
This guide is about the second problem: how to wire forms, logins and CRM imports into one identity record where every person-level match carries a consent record — and what to do when the record is missing.
What counts as first-party data in B2B
The three sources you already have
- Forms. Demo requests, gated content, event signups. Capture the form ID, page URL, the text shown next to the submit button and the timestamp.
- Logins and accounts. Signup, passwordless login, invited seats. An authenticated session is the strongest identity claim you get, because the person proved control of an inbox or a passkey.
- CRM records. Contacts and companies created by sales, imported lists with a documented source. An import is first-party only if you can name where the data came from.
What stays anonymous by design
A visit from home broadband, a mobile carrier, a hosting provider or a VPN does not name a company. Reveal names a company only when the visit comes from that company’s own network, and shows the confidence and evidence behind the match. That boundary is why the company-level data stays defensible.
Build the consent record before the match
Most consent records get rebuilt later from a form export. Build them at write time: when an identifier enters your system, attach the record and never edit it afterwards.

The fields a usable consent record needs
{
"email_hash": "sha256:...",
"source": "form:demo_request",
"page_url": "https://yourdomain.com/demo",
"notice_version": "2026-05-01",
"consent_text": "text shown, verbatim",
"collected_at": "2026-05-14T09:31:07Z",
"policy_tag": "marketing_contact",
"account_id": "acc_1842"
}
Add the collection method, the workspace and a version of the record itself. Store it append-only, export it on request, and join it to the person identifier at insert time. If your tools hold separate copies, decide which one is authoritative.
Where teams break it
Three failures show up repeatedly: matching on an email domain that is a consumer mailbox, adding matches onto contacts who never submitted anything, and treating a purchased list as first-party.
Match person to company without guessing
Rules that hold up
Match a person only when the record carries an identifier they gave you: the email they submitted, the account they logged into, a CRM entry with a documented source. Keep the evidence next to the match — which form, which timestamp, which notice version — and prefer the newer authenticated source when two records conflict.
Deliver to the CRM as an event, not a rewrite
Push a match as an event with the consent record attached instead of overwriting contact fields your sales team maintains by hand. Let the CRM stay the system of record for people; let the identification layer write the evidence into it. Reveal delivers to Slack, HubSpot, Salesforce, n8n, Zapier and signed webhooks.
Logins are the cleanest signal you have
An authenticated user is not an anonymous visitor with a better score. When someone logs in you know the account, the seat and often the plan. Attach the consent record at signup, keep the account ID on every later event, and treat product activity as intent rather than as a new identity to resolve.

Forms need the same care. Scripted submissions, disposable addresses and OTP pumping arrive through the fields you rely on for consent, so form-level protection and identity resolution should share one policy. Our form protection notes cover the decision layer that runs before a submission is accepted.
Keep the record honest after the match
People withdraw consent, change roles, move companies and ask what you hold about them — so the record has to work after the match, not only at the moment of capture.
Publish an opt-out path that does not require an account, keep a suppression list, and make sure a suppression removes future person-level matches rather than hiding them on one screen. The Reveal privacy notice explains who processes data on your behalf, and the opt-out page is public.
Then run a quarterly check: pick ten matched people and have a colleague reconstruct each origin from the stored record alone. If any of them needs an explanation from the person who built the pipeline, the record is not finished.
FAQ

Is first-party data enough to contact someone?
It depends on your jurisdiction and lawful basis, which is a question for your counsel. What you control is evidence: if the record shows the form, the text displayed, the timestamp and the notice version, your team can answer quickly.
Can you identify a company without identifying a person?
Yes. Company-level identification works from the network the visit comes from, with confidence and supporting evidence, and stays anonymous when the visit comes from home broadband, mobile carriers, hosting or a VPN. People are linked only from first-party data.
Where should the consent record live?
Keep one authoritative copy and a reference in the other. What matters is that the record travels with the person identifier, so a match can never appear without it.
See it in practice. Reveal names companies from their own network, scores intent with evidence a rep can read, and links people only from data they gave you — forms, logins and CRM imports — with a consent record behind every match. Start free at pryntid.com.
Try it free
Prynt is device intelligence with a free plan: 5,000 identifications a month, visitor IDs with confidence and bot detection — no credit card. Upgrade to Pro for VPN, proxy, tampering and behavioral signals.