When Intent Signals Are Wrong: False Positives in Account Prioritization

This is not the GDPR piece and not the US-versus-Europe piece. Those cover whether you're legally allowed to track a signal and how buying committees behave once you have one; this one covers a different question entirely: whether the signal was ever real.

An intent platform flags an account: a pricing-page visit, two whitepaper downloads, a surge on "B2B marketing automation" inside 48 hours. Sales calls. The account turns out to be a product manager at a competitor running a quarterly feature audit and, separately, a master's student building a comparison matrix for a dissertation chapter. Nobody finds out until the SDR has already burned a touch sequence on an account that was never going to buy anything.

That's not a tooling failure. The platform did exactly what it was built to do: it saw behavior that looks like buying intent and reported it. The failure sits upstream, in the assumption that behavior resembling intent is intent. It usually isn't, and the gap is wide enough that most teams running on unfiltered intent data are routing real noise straight into their sales pipeline without knowing it.

TL;DR: Intent signals fail in predictable, nameable ways: competitor and vendor research, students and job seekers, review-site traffic, IP-to-company resolution errors on remote and VPN connections, and topic-surge noise from data co-ops. Each has a specific mechanism and a specific fix, not a vague "add more data sources." You can measure your own false-positive rate by auditing closed-lost and no-opportunity accounts that were intent-flagged, checking your IP resolution error rate directly, and tracking how often SDRs override or ignore a "hot" account. Once you have that number, the fix is a layered validation gate before anything reaches a rep: cross-reference signal types, require first-party corroboration, set a minimum firmographic fit threshold, and QA a sample by hand every cycle. Skip the gate and you train your own sales team to stop believing the data, which is a harder problem to undo than the noise itself.

What a False Positive Actually Is Here

A false positive in account prioritization isn't a tracking error. The visitor was real, the pageview happened, the IP resolved to something. The error sits in the interpretation layer: a system reads "this account is in-market" when the correct read is "someone at or near this account did something that people who never buy also do." A reverse-IP hit, a topic surge, a review-page visit are all real events. None is, on its own, evidence of a buying process.

The cost shows up downstream, not at the point of detection. A rep works a flagged account, gets nothing, and the next flagged account gets a slower, more skeptical response. Sabnis, Chatterjee, Grewal and Lilien's study "The Sales Lead Black Hole: On Sales Reps' Follow-Up of Marketing Leads," published in the Journal of Marketing in 2013, is built around this exact dynamic: reps allocate follow-up effort based on their own running estimate of lead quality, and that estimate updates fast when the leads they're given keep turning out to be duds. One wrong signal is a wasted call. A signal source that's wrong often enough becomes a source nobody on the floor opens anymore, regardless of how good the next signal actually is.

Where False Positives Actually Come From

Six mechanisms account for most of the noise in a typical intent stack, each with a distinct cause, which is why "add a filter" is really six separate fixes stacked together.

The IP-resolution row is the least visible mechanism to a marketing team that doesn't work with networking data day to day. Reverse-IP identification matches a visitor's IP address to a company through WHOIS and ISP records, a method built for an era when B2B traffic came from a stable office IP block. Gallup's 2025 tracking of U.S. hybrid work found the hybrid-work retreat many executives expected hasn't materialized, so a large share of the knowledge workers your platform is trying to identify still connect from home ISPs or VPNs instead of one corporate network block. Knock2 covers this directly in "Why VPNs and Corporate Networks Break Website Visitor ID Match Rates," and the underlying geolocation problem is old and well documented: a 2017 Internet Measurement Conference study, "A Look at Router Geolocation in Public and Commercial Databases," tested commercial geolocation databases against verified ground-truth router locations and found they routinely disagreed with each other and with reality, a finding APNIC covered for a networking audience. Even MaxMind, one of the larger geolocation vendors, publishes its own accuracy comparison showing material variance by country instead of one blanket claim. None of this makes IP-based identification useless. It makes it one signal with a documented error rate, not a verdict.

Review-site noise compounds the problem rather than replacing it. G2's own newsroom announced that it had "expanded buyer intent across four software discovery platforms, delivering up to 2x more signals," which is good news for signal volume and bad news for anyone without a filter, since more signal surface from a platform that competitors and researchers use the same way buyers do means more raw events, not more qualified accounts.

How to Quantify Your Own False-Positive Problem

Most teams never measure this because "is our intent data wrong" sounds unanswerable from the inside. It isn't. Four checks against your own CRM history will give you a real number within a week.

Pull every account flagged high-priority in the last two quarters and join it against closed-lost and no-opportunity outcomes. An account that was "hot" by intent score and never produced a single qualified conversation is a false positive or a process failure; sampling 30 to 50 by hand tells you which. Then check your IP resolution error rate directly rather than trusting a vendor's blended claim. Knock2's own guidance, "How to Vet a Website Visitor ID Vendor's Accuracy Claims," makes the point that a vendor's aggregate number hides wide variance by traffic source, company size, and geography; you want your error rate on your traffic, not theirs on a reference panel.

Segment flagged accounts by source next. A fixed weekly share of your "intent-qualified" list coming from review-platform referrals or a narrow set of academic and competitor domains is a structural leak, not bad luck. Last, and most revealing, track your SDR override rate: how often a rep disqualifies a flagged account on the first call, and why. A rate that climbs quarter over quarter on one source means your reps already ran the audit informally. The data just hasn't reached whoever configures the scoring model.

The Validation Gate: What Should Happen Before a Signal Reaches Sales

A single intent signal should never be enough to route an account to a rep. The fix isn't a smarter algorithm; it's a gate with more than one door.

Purple path's framework for sequencing these checks starts with firmographic fit before behavioral signal: a perfectly matched behavioral signal on a company with ten employees and no budget is still a bad account, while a company that fits your ICP but hasn't shown behavior yet is a nurture problem, not a disqualification. The full reasoning is in "Firmographic vs. Behavioral ICP Signals: Which Should Come First," and the matching step behind the account-fit gate is covered in how to build a TAM list your intent platform can actually check accounts against. Once a signal clears every gate, the handoff needs a defined path into the CRM and sequencing tools your reps already use, covered separately in how to integrate intent data into CRM and automation workflows.

Why This Is a Different Question Than GDPR or Regional Buying Behavior

These three topics sit close together and get conflated often, so the boundary is worth drawing explicitly. Legal permission to collect and act on a given signal under European data protection law is a compliance question, and purple path has covered it directly in "Intent Data and GDPR: What European B2B SaaS Companies Can Legally Track." How differently a European buying committee responds to a given signal compared to a US one is a behavioral and cultural question, covered in purple path's piece on why intent data behaves differently in European buying committees than in the US. Neither question touches whether the signal itself was accurate in the first place. A perfectly legal, perfectly culturally-calibrated signal that's actually a competitor's product manager running a feature audit is still a false positive, and no amount of compliance or regional tuning fixes that.

Frequently Asked Questions

What's a realistic false-positive rate to expect from intent data?

No single industry-wide number is worth quoting; treat any source claiming one blended figure across every vendor and traffic mix with skepticism. Measure your own rate with the closed-lost audit and SDR override tracking above, since it depends on your traffic mix, review-site visibility, and remote-work share.

Is IP-to-company resolution getting better or worse over time?

Both at once. The technology keeps improving, but Gallup's continued tracking of U.S. hybrid work shows no significant retreat from remote arrangements, so the VPN and home-ISP mismatch problem isn't shrinking as fast as the tooling.

Should we just turn off review-site intent data entirely?

No. It's a legitimate corroborating signal, not a reliable standalone trigger. Demote it to a secondary signal that has to agree with at least one other signal type before an account gets routed, rather than losing the genuine buyer behavior it does capture.

Who should own the false-positive audit: marketing, sales ops, or RevOps?

Whoever owns the intent-to-CRM pipeline, because the fix spans scoring logic, CRM routing, and SDR process. In practice that's usually RevOps, with sales ops supplying override-rate data and marketing supplying the signal-source breakdown.

How often should the validation gate be reviewed?

Quarterly at minimum, and immediately after any change to your intent vendor, your ICP definition, or a major market shift: a new well-funded competitor, a layoff wave in your target industry, a hiring surge that changes your job-seeker noise profile. A filter built once and never revisited is how last year's suppression list becomes this year's blind spot.

Getting the Gate Built Right

A validation framework is only as good as the discipline behind maintaining it, which is the part most teams underinvest in once the initial build is done. If your pipeline routes flagged accounts to sales faster than anyone audits whether those accounts were ever real, that's a go-to-market design problem, not a data problem, and it's the kind of thing purple path builds for B2B SaaS teams as a matter of course. Learn more about how purple path approaches go-to-market and account-based programs and get in touch if your sales team has started quietly ignoring your "hot" accounts.