Most B2B teams treat every visit as equal until they export a list and see the problem. The pages that signal real buying intent are rarely the ones with the most traffic. Scoring high intent pages B2B is a weighting exercise, not a volume exercise: one pricing visit from a company that matches your ICP tells you more than forty blog reads from a research firm that will never buy.
Reveal, the visitor identification product in the Prynt suite, names the company behind a visit when the visit comes from that company’s own network, attaches an explainable intent score, and delivers the account to your CRM or Slack. Most teams get the weighting wrong by adding more pages instead of asking what each page answers.
One boundary first: home broadband, mobile carriers, hosting and VPN visits stay anonymous in Reveal. A score built on anonymous noise is worse than no score.
Score the question, not the page view
Every page answers a question in the buyer’s evaluation, and the question is what carries intent. A visitor reading a definition is orienting; a visitor comparing plan limits at 11 p.m. is deciding.
Assign each page template a band — high, medium, low — instead of a raw multiplier per URL. Bands survive redesigns and new landing pages; per-URL scores rot in a quarter and nobody remembers why an old pricing path had a weight of nine.
Two rules keep bands honest. A page earns a high band only if a real evaluation step happens on it. And every high band must be explainable in one sentence, because your sales team will ask.
The four page types that carry the most intent

Pricing and plan pages
Pricing is the clearest evaluation signal on most B2B sites, but depth matters more than arrival. A four-second bounce differs from a scroll to the enterprise column, a click into plan comparison and a return visit three days later. Score arrival medium, a repeated deep visit high.
Docs, API references and changelogs
Documentation visits mean someone technical is testing feasibility: authentication, rate limits, response shapes, whether the product does what the landing page claims. Strong for developer-led products, weak for marketing-led ones. Weight it high only for product-specific docs paths, not the general help center.
Integrations and stack pages
An integrations page is a compatibility check, usually read after the buyer accepts the category matters and is now asking whether you fit the stack they already run. Weight specific integration pages high, directory pages medium.
Security, privacy and compliance pages
These pages are read by the person who can veto the purchase. A visit here often means procurement, legal or IT has entered the conversation — later and more serious than an early demo request. Weight it high, and treat security plus pricing in one session as a strong combined signal.
Weight without overfitting
Overfitting looks like a score that hits 95 after a single page view. Four habits keep it honest:
- Cap each page type. One template contributes a maximum, however many of its URLs get visited. Otherwise a docs crawl becomes a hot lead.
- Decay by recency. A visit this week should outweigh the same page from two months ago. Pick a window and keep it constant.
- Require two signals for the top band. Pricing plus docs, or docs plus security. Single-page top scores are usually bots, internal QA or a competitor’s researcher.
- Exclude what is not a buyer. Internal traffic, bot sessions, agency partners. Prynt Identify flags bot and AI-agent traffic; Reveal only names companies visiting from their own network.
Then test the model against reality. Pull your last twenty closed-won accounts, note which pages they touched before the deal, and compare with twenty closed-lost ones. If lost deals rank as hot, the weights measure curiosity, not intent. Our accuracy page explains how evidence is reported per match.
Gate the score with ICP fit
Intent without fit is a calendar full of bad demos. A fintech reading your pricing page is hot only if you sell to fintech. Keep fit and intent as separate fields: high fit plus high intent goes to sales today, high fit plus low intent to nurture, low fit plus high intent to a watch list.

Reveal computes ICP fit and segments next to the intent score, so the record shows both numbers and the reasons behind them. A rep can argue with a reason; they cannot argue with a single number.
Put the weights in a playbook, then review monthly
A score nobody acts on is a dashboard. Turn it into rule-based playbooks: an account crossing the threshold lands in Slack with the pages it read, is updated in HubSpot or Salesforce, and gets a drafted message a person reviews and sends. Nothing leaves automatically.
Review monthly, not weekly. Look at accounts that scored high and never replied, accounts that converted from a low score, and any template whose band you cannot justify. Change one band at a time so you can tell what moved. Our Reveal privacy notice covers how visits are matched and how opt-outs are honored.
FAQ

Does one pricing page visit mean buying intent?
No — it means evaluation started. Treat a single pricing visit as medium intent and require a second signal (docs, integrations, security or a return visit) before escalating to a person.
How long should a page’s weight last?
Long enough to cover a normal evaluation cycle, short enough that last quarter’s research does not haunt the score. A 30- to 90-day window works for most teams: pick one, write it down, keep it fixed while you test other variables.
Can I see the person, or only the company?
Reveal names the company from its own network. People are linked only from your first-party data — form fills, logins, CRM imports — with a consent record attached. Anonymous visitors stay anonymous; no data brokers, no guessing.
Where to start
Pick four page templates, assign bands, cap each one, and run the model against last quarter’s deals before wiring it to Slack. If the ranking holds, ship it. If not, the audit taught you more than another month of pageviews.
Start with Reveal on Prynt, and keep our docs open while you set thresholds.
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.