How to Choose B2B Website Visitor Identification Software: A 14-Day Buying Guide

To choose B2B website visitor identification software, decide first whether you need companies or people, then run two shortlisted tools on the same pages for 14 days and score them on the visitors you can verify, not on the rate the vendor quotes. This buying guide gives you the decision criteria, a scoring worksheet you can copy, a day-by-day test plan with real numbers, and the red flags to catch in vendor claims before you sign.

It does not rank vendors. For the tool-by-tool shortlist with October 2026 prices and review ratings, see our comparison of website visitor identification software.

Step 1: Decide What Job You Are Hiring the Tool For

Most bad purchases in this category come from buying the wrong level of identification, not the wrong vendor. Answer three questions before you book a single demo.

1. Companies or people? Company-level identification names the business behind a visit and works worldwide. Person-level identification names the individual (name, title, LinkedIn) and in practice only works on US traffic.
2. What share of your traffic is American? Pull the last 90 days from your analytics. If US visitors are under 20% of your traffic, person-level features will produce a handful of names a month, whatever the vendor's coverage claim.
3. Who acts on the output, and where? An SDR team working Slack alerts needs speed and contact data. A marketing team building account lists needs CRM sync and clean company records. A founder checking pipeline needs a weekly digest. The workflow decides which integrations are must-haves.

Write the answers down. They set the weights in the worksheet below.

Step 2: The Decision Criteria That Matter

These are the criteria that separate tools once you test them on real traffic. Feature checklists from vendor sites tend to list everything; these are the ones that change outcomes.

• Verified coverage of your known visitors. Of the accounts you know visited (demo bookings, replies, sales calls), how many did the tool catch?
• Junk rate. What share of "identified companies" are internet providers, hosting networks, universities or bots? Every junk row costs review time, and on some pricing models it also costs credits.
• Match quality. Is the company the right one, not a parent group or a coworking provider? For person-level matches, are the title and LinkedIn profile current?
• Pricing shape at your volume. Price the plan on the number of new companies and people you will identify each month, not on your visits. Then check what happens at twice that volume.
• Integration depth. Native CRM sync with field mapping and deduplication, or CSV and webhooks only? Where in the plan ladder does it start?
• Workflow fit. Alerts on the pages that matter (pricing, demo, case studies), filters by fit, and a behavioral score your team will actually read.
• Compliance posture. A consent mode for EU traffic, support for opt-out signals, and a clear statement that EU individuals are not identified without consent.

Step 3: The Scoring Worksheet

Score each tool 1 to 5 on each criterion at the end of the test, multiply by the weight, and add the totals. The default weights below suit a team that sells to mid-market accounts with mixed US and European traffic.

Adjust the weights to your motion. An SDR-led US team should move 10 points from integration depth to match quality. An ABM team running account lists should move 10 points from workflow fit to integration depth. A team with mostly EU traffic should drop person-level features from the test entirely and score company coverage only.

A worked example: Tool A scores 4, 3, 4, 5, 2, 4, 5 and Tool B scores 3, 4, 3, 3, 5, 3, 5 on the seven rows. With the default weights, A totals 3.75 and B totals 3.55. The tie-breaker is the CRM row: if your team will not work from CSV exports, B's integration score outweighs A's coverage lead, and you should raise that weight before you add up the totals, not after.

Step 4: The 14-Day Test Plan

Fourteen days of normal traffic is the practical minimum: long enough to catch return visits and a full two-week sales cycle of weekdays, short enough to keep the decision moving. Run both tools at the same time on the same pages, so traffic mix cannot explain the difference.

Day 0: set up and write down your ground truth.

• Install both scripts site-wide. They run independently and do not interfere with each other.
• Build a ground-truth list of 15 to 30 accounts you expect to visit in the next two weeks: open opportunities, recent demo bookings, prospects your SDRs are emailing, customers in onboarding.
• Record your baseline: visits per day, US share and top pages from your analytics.

Days 1 to 7: let it run, change nothing.

• Do not tune filters or alerts in week one. You are measuring the default experience.
• Export each tool's identified companies at the end of day 7. Free plans with short history windows (7 days on some tiers) will drop older rows, so export weekly.

Day 7: midpoint check.

• Count identified companies per tool and the share that are junk.
• Mark which ground-truth accounts each tool has caught so far.
• Now set up the one workflow you would run in production, such as an alert on pricing-page visits from target accounts, in both tools.

Days 8 to 14: test the workflow.

• Act on alerts the way your team would. Count how many led to a real follow-up.
• Push at least one record to your CRM from each tool, through whatever path the plan you would buy supports.

Day 14: score.

• Fill in the worksheet. Spot-check 20 random rows per tool by hand for match quality.
• Price month one: take the number of new companies and people each tool identified in 14 days, double it, and price that on each vendor's plans.

What to expect in numbers. Use our measured rates as a sanity check, not a target. Across 14 live sites (July 27 to October 23, 2026), 21.5% of visits and 12.7% of unique visitors matched to a company, and 12.1% of unique US visitors matched to a person. Per-site company rates ran from 0.7% to 68.6%, with a median of 18.5% across sites with at least 50 visits. The method is on our match rates page.

So for a site with 2,000 unique visitors in the 14 days and a 25% US share, a result near 250 identified companies and 60 named US visitors would be typical. A result of 40 companies is not necessarily a broken tool; it can mean your audience works from home. That is exactly why the ground-truth list matters more than the raw count.

Mind the free-plan limits. Free tiers are built for a first look, not a 14-day test. VisiLead's free plan has 10 one-time credits (one per new company or person), which a site with 100 visits a day can use up on the first day at our median match rate; budget one month of a paid plan ($29 for Starter) if you want the full 14 days. Other vendors cap free plans by companies shown, history length or person-level data, so read the limits before day 0 or one tool will look worse simply because it stopped counting.

Low traffic? If your site gets fewer than 500 visits in two weeks, run the test for 28 days and weight the ground-truth row even higher, because the raw counts will be too small to compare.

Step 5: Red Flags in Vendor Claims

Vendor match-rate claims range from "15-20% coverage" to "over 75% of companies", and they rarely count the same thing. Our comparison post lays the published claims next to measured numbers. When a vendor quotes you a rate, check for these.

• No denominator. A rate without "of visits", "of unique visitors" or "of companies" cannot be compared to anything. Ask in writing.
• "Up to" rates. An upper bound measured on the friendliest site in the customer base tells you nothing about yours.
• Person-level rates on total traffic. Person-level identification is a US product. A rate quoted on worldwide traffic for EU-heavy sites should raise questions about how it is achieved.
• EU individuals identified without consent. Any vendor that says it names EU visitors by default is either counting company matches as people or taking a legal risk you will share.
• Junk counted as matches. If internet providers and cloud hosts appear as "companies", the headline rate is padded. Ask whether those rows count against your quota.
• Emails for every named visitor. Ask where the email comes from, whether it is a work or personal address, and what share of person matches include one.
• Monthly prices that bill annually. "€79 a month" billed as one annual invoice is a different commitment from a monthly plan you can cancel.
• Price only after the trial. Quote-based pricing set on your traffic during a trial puts the vendor in control of the number. Ask for the price range before you install.

Questions to Ask Every Vendor in Writing

1. "What match rate should we expect on our traffic mix, and what is the denominator?" A good answer gives a range and asks about your US share.
2. "Which networks do you filter out, and do filtered or junk matches cost credits?" A good answer names internet providers, hosting networks and bots, and says no.
3. "Can you identify EU visitors at the person level?" The only defensible answer is no, not without consent.
4. "Where does CRM sync start, and what does it map?" Get the plan name and the fields.
5. "What happens to our data if we cancel?" Export path and deletion timeline should be immediate answers.

For the legal side, our guide to GDPR-compliant visitor identification explains legitimate interest, consent and retention. For the level question, see person-level vs company-level visitor identification.

Decision Rules After the Test

• If one tool caught clearly more of your ground-truth accounts, it wins unless its junk rate or CRM path makes the output unusable.
• If the tools tie on coverage, decide on pricing at your volume and on the workflow your team actually used during the test.
• If neither caught more than a fifth of your ground-truth list, the problem is likely your traffic mix, not the vendor. Spend on the cheaper tool, or put the budget into channels that bring office traffic.
• Whatever you pick, re-run the ground-truth check every quarter. Data quality drifts, and anything built on top, such as your lead scoring model or a CRM handoff, is only as good as the identification feeding it.

Frequently Asked Questions

Q: How do I choose B2B website visitor identification software? A: Decide whether you need companies or people, check your US traffic share, then run two shortlisted tools on the same pages for 14 days. Score them on how many of your known visitors each caught, junk rate, match quality, pricing at your volume, CRM integration and compliance. Choose on the score, not the vendor's quoted rate.

Q: What is a good match rate for visitor identification? A: There is no universal number. On VisiLead's sample of 14 sites, 21.5% of visits and 12.7% of unique visitors matched to a company, and 12.1% of US visitors to a person, but individual sites ranged from 0.7% to 68.6%. A better test is how many of your known visiting accounts each tool caught.

Q: How long should a visitor identification trial run? A: Two weeks of normal traffic is the practical minimum, because it captures return visits and a full cycle of weekdays. If your site gets fewer than 500 visits in two weeks, run four weeks instead. Check the free plan limits first: some cap companies, history or credits and will stop counting before the test ends.

Q: Should I buy company-level or person-level identification? A: Start from your traffic. Person-level identification only works reliably on US visitors, so it pays off for teams with substantial US traffic and an outbound motion. Company-level identification works worldwide and suits account-based sales and marketing. Many teams start with companies and add person-level when US outbound becomes a core channel.

Q: What are the red flags in visitor identification vendor claims? A: Watch for match rates with no denominator, "up to" figures, person-level rates quoted on worldwide traffic, claims to name EU individuals without consent, junk networks counted as companies, and monthly prices that are billed annually. Ask each vendor to put its expected rate and its definition in writing before you start a test.