A founder at a Series A dev tools company showed me his analytics last month. 22,000 visits the prior quarter. Fourteen demo requests. He wanted to know why his paid ads were "broken."
The ads were fine. The problem was the 21,986 people who read his docs, priced out a plan, compared him to a competitor, and then closed the tab. He had no idea who any of them were. They looked at buying, decided something, and left no trace he could act on.
That gap is the single biggest wasted asset in most B2B go-to-market motions. You spend real money getting the right accounts to your site. Then you let almost all of them walk out the back door because they did not fill in a form. Warm outbound is the practice of closing that door: identifying who was on your site, deciding which visits actually mean something, and reaching out with a message tied to what they looked at.
I have built this motion for about a dozen companies now, first as a RevOps consultant and now as Founding GTM Engineer at Peec AI. It works. It also gets sold to founders in a way that is close to dishonest, so this post is the version with the math attached. What identification actually delivers, what it does not, the numbers you should expect, and how to build the system around it so it produces meetings instead of a spreadsheet nobody opens.
Roughly the share of B2B website visitors who leave without filling in a form. For most companies that traffic is invisible and unworked.
What website visitor identification actually is
Strip away the marketing and there are two very different things being sold under one name.
Company-level identification resolves a visitor's IP address or first-party cookie to a business. You learn that someone from Acme Corp looked at your pricing page twice this week. You do not learn who. This is the older, more reliable, more legally settled version. Tools like Leadfeeder, Factors, and Clearbit Reveal (now Breeze Intelligence inside HubSpot) live here.
Person-level identification goes further and tries to name the individual. First name, last name, LinkedIn profile, sometimes a work email. This is what made RB2B blow up in 2024 and 2025. It works by matching your anonymous traffic against a third-party identity graph built from cookie pools and data partnerships. When it hits, it is genuinely useful. When it misses, you get nothing.
The distinction matters more than any feature comparison, because the two approaches have completely different match rates, different price tags, and very different legal footing. Most founders I talk to do not know which one they bought.
The match-rate numbers nobody puts on the pricing page
Here is the part vendors bury. A "90% match rate" almost always means company-level, on US traffic, counting only the visits the tool decided to attempt.
The honest ranges, from independent tests and my own implementations:
Company-level lands somewhere between 30 and 70 percent depending on how much of your traffic is real companies versus consumer ISPs, VPNs, and mobile. Person-level is the one people get burned on. Vendor decks quote 60 to 75 percent. Real reviewers report 8 to 35 percent once you account for ad blockers, Safari, incognito, and the fact that the identity graph mostly covers the United States.
So if you run a company selling into Germany, France, the UK, or the Netherlands, person-level tools like RB2B will identify almost none of your traffic. This is not a bug you can configure around. The identity graph does not have European coverage, and European privacy law makes building one a legal minefield. I will come back to that.
The math test I give every founder before they buy anything: take your monthly visitor count, multiply by a realistic person-level match rate of 15 percent, then multiply by a 3 percent visitor-to-meeting rate on cold-ish warm outreach. A site with 5,000 monthly visitors gets you about 22 identified people and maybe two meetings a month. Below roughly 5,000 monthly B2B visitors, the tooling cost is higher than the pipeline it produces. Do not buy it yet. Fix your traffic first.
Why identification is the easy 10 percent
Here is the thing most people get wrong. They treat the tool as the project. Install the pixel, connect Slack, done. Six weeks later the Slack channel is a firehose of names nobody actions and the pilot gets quietly canceled.
Identification is maybe 10 percent of the work. It tells you who showed up. It does not tell you which of those visits mean anything, who should own the follow-up, what to say, or how fast. That system around the data is the 90 percent, and it is the part no vendor builds for you.
Think about what a raw feed actually contains. Your own employees. Your existing customers checking the docs. Competitors doing research. Job seekers reading your careers page. Someone who bounced off the homepage in four seconds. Buried in there are maybe 5 percent of visits that represent a real buyer moving closer to a decision. Send generic "I saw you visited our site" messages to the other 95 percent and you have built a spam machine that also happens to be creepy.
The difference between these two is not the tool. Both use the same pixel. The difference is the filtering, routing, and messaging logic you build on top. That is where the pipeline comes from.
The warm outbound system, step by step
This is the architecture I build. Five stages, and the tool only handles the first one.
Step 2: qualify before a human ever sees it
The first filter kills most of the noise. Run every identified account against your ideal customer profile automatically. Wrong size, wrong region, wrong industry, existing customer, existing open opportunity, competitor domain: all of these get suppressed before they reach a rep. I usually do this in Clay or a lightweight n8n workflow that reads the visitor feed, enriches the company, checks it against CRM, and only passes clean records forward.
This one step is the difference between a tool people trust and a tool people mute. If the first ten alerts a rep sees are garbage, you have lost them, and you will not get their attention back.
Step 3: score by page, not by visit count
Not all pages carry the same signal. Someone who read one blog post is curious. Someone who hit the pricing page, then the security page, then a competitor comparison, inside two days, is building an internal case. Score accordingly.
My rough weighting, and you should tune it to your own funnel: pricing page is worth a lot, comparison and competitor pages a lot, docs and integration pages a fair amount, product feature pages some, top-of-funnel blog almost nothing on its own. Two or more high-value pages in a short window is the threshold that actually correlates with a reply. This is first-party intent, and it is the most reliable signal you will ever get. I wrote a whole piece on why in the B2B intent data guide.
Step 4: route by fit, not by round robin
A named person at a 2,000-person target account who priced out enterprise is not the same as a company-level match at a 15-person shop reading your blog. They should not get the same treatment. High-fit, high-intent goes straight to the owning AE with a Slack alert and enough context to act in five minutes. Mid-fit goes to a lighter sequence. Everything else goes to nurture or to nothing at all. Doing nothing is a valid, often correct, routing decision.
Step 5: message the problem, not the pixel
The fastest way to torch a warm outbound program is to open with "I saw you were on our website." It is creepy, it advertises that you are tracking people, and it leads with your surveillance instead of their problem. Do not do it.
Instead, use the page they looked at as private intelligence about what they care about, then write an email that speaks to that problem as if you inferred it from context. Someone deep in your security and compliance pages probably has a SOC 2 or data residency question. Someone comparing you to a named competitor is in an active evaluation. Write to the problem. The relevance lands without you ever revealing how you knew.
The GDPR reality most of these articles skip
Almost every article on this topic is written from a US point of view, which is a problem if you sell into Europe. Here is the honest version.
Company-level identification is generally fine in a B2B context. Resolving an IP to a business name, without identifying a person, can usually run on legitimate interest as your legal basis. Most European DPAs treat this as acceptable when you are transparent about it in your privacy notice. This is why company-level tools work across regions and person-level ones mostly do not.
Person-level identification of an EU or UK resident is a different animal. An IP address can be personal data under GDPR, and naming an individual from anonymous traffic is personal data processing that needs a lawful basis. In practice that means explicit consent through a compliant cookie banner, with the identification cookie classified as non-essential marketing. Almost nobody visiting your site consents to that, so the effective match rate on EU traffic is close to zero, and the legal exposure if you do it without consent is real.
My advice for European companies, or anyone with meaningful EU traffic: run company-level identification and build your whole warm outbound motion on that. You lose the individual name but keep the account signal, which is 80 percent of the value with a fraction of the risk. For US traffic you can layer person-level on top. Do not run a US person-level tool against your German pipeline and assume it is someone else's compliance problem. It is yours.
The tools, honestly
I will not do a 15-row comparison table, because your choice comes down to two questions: which regions is your traffic in, and how much of it is there.
For US-heavy traffic where person-level matters, RB2B has transparent pricing and a LinkedIn-first workflow. Warmly bundles identification with orchestration and AI outreach if you want more of the system in one place, though you pay for it. Vector is strong if you want deep CRM and Slack plumbing. For company-level across regions, Factors, Leadfeeder, and Clearbit Reveal inside HubSpot are all reasonable, and if you already run HubSpot the native Breeze Intelligence option removes an integration.
But I would not start with the tool. I would start with the traffic math and the system design, then pick the cheapest tool that covers your regions, and put the saved budget into the routing and enrichment layer where the pipeline actually gets made. This is the same principle behind good CRM and RevOps architecture: the connective tissue matters more than any single platform.
Sitting on traffic you are not working?
Book a free 30-minute audit and we will map the warm outbound system for your stack, region, and traffic volume, and tell you honestly if you are big enough to bother yet.
Book an audit →What good looks like after 90 days
When this is built right, the outputs are boring and measurable. Reps get a handful of high-quality alerts a day instead of a firehose. AEs act on the good ones within an hour because the context is already attached. A meaningful slice of your pipeline, often 10 to 20 percent within a quarter, starts carrying a "sourced from warm outbound" tag. And the reply rates look nothing like cold outbound, because the person on the other end was already looking.
That last point is the whole game. You are not interrupting a stranger. You are following up with someone who raised their hand quietly and assumed nobody noticed. Notice, filter hard, and reach out about their problem. That is warm outbound. Everything else sold under the name is a pixel and a prayer.
If you want help wiring the automation layer that does the qualifying and routing, that is exactly the kind of build we do.
Frequently asked questions
What is warm outbound?
Warm outbound is reaching out to prospects who have already shown interest in your company, usually by visiting your website, before they fill in a form. You identify the anonymous visitor, decide whether the visit signals real buying intent, and follow up with a message tied to what they looked at. It sits between cold outbound, where the prospect has no prior contact, and inbound, where they self-identify.
How accurate is website visitor identification?
Company-level identification, which names the business but not the person, matches 30 to 70 percent of US B2B traffic. Person-level identification, which names the individual, realistically matches 5 to 20 percent and works mostly on US traffic. Vendor claims of 80 to 90 percent almost always refer to company-level matching on filtered US visits, not people.
Is website visitor identification legal under GDPR?
Company-level identification is generally acceptable in a B2B context on a legitimate interest basis, as long as you disclose it in your privacy notice. Person-level identification of EU or UK residents needs explicit consent through a compliant cookie banner, which almost nobody grants, so it rarely works on European traffic and carries real legal risk if done without consent.
How much website traffic do I need for this to work?
Below roughly 5,000 monthly B2B visitors, the tooling cost is usually higher than the pipeline it produces. At 15 percent person-level match and a 3 percent visitor-to-meeting rate, 5,000 visitors yields around two meetings a month. If your traffic is lower, fix demand generation first before buying an identification tool.
What tools do I need for warm outbound?
You need an identification tool that covers your regions (RB2B or Warmly for US person-level, Factors or Clearbit Reveal for company-level across regions), an enrichment and routing layer (Clay or n8n) to filter out bad-fit accounts, and your CRM to check for existing customers and open deals. The identification tool is the smallest part. The filtering and routing logic is where the pipeline comes from.
Stop letting buyers leave anonymous
Most companies spend heavily to get the right accounts to their site, then work less than 5 percent of the people who show up. The fix is not a bigger ad budget. It is a system that catches the buyers who were already there, filters hard, and reaches out about the problem they came to solve.
If you want that built on your stack, in your regions, without the legal exposure, talk to us. We will tell you what to turn on, what to skip, and whether you are ready for it at all.