Negative ICP Signals: How to Disqualify Bad-Fit PLG Signups Before They Reach Sales

The scoring model nobody builds

Most ICP scoring in PLG only runs in one direction. You score someone up: right title, right company size, right industry, and their number climbs. Rarely does anyone build the other half — the list of signals that should pull a signup's score down or kill it outright before a sales rep ever sees the name.

That gap matters more than it looks. A free-tier SaaS tool typically pulls in students doing a class project, someone's personal side project, a competitor's product team poking around, and a handful of consultants testing the waters for a client who isn't even in the room. None of them are bad people. None of them are your buyer either. If your ICP scoring model only measures fit going up, all of that noise still lands in the same queue as a VP of Growth at a 200-person Series B company.

Negative ICP signals fix that. They're the disqualifying criteria your model should check first, before it bothers computing a positive score at all.

Why positive-only scoring breaks down

A positive-only model treats "we don't know" the same as "weak fit." Someone with a personal Gmail domain and a vague job title doesn't score high, but they don't score zero either — they just sit in a gray middle tier where a rep has to manually look them up to decide if they're worth a call.

That manual look-up is the real cost. Every signup a rep has to research by hand is time not spent on the handful of signups that are obviously worth a conversation. In a PQL triage queue, the gray middle tier is usually the largest bucket, and it's the one that burns the most rep hours per qualified meeting booked.

Disqualification signals shrink that middle tier directly. Instead of asking "how good is this fit," the first pass asks "is there a clear reason this signup can't convert, regardless of fit." If yes, route it out of the sales queue immediately. If no, then run the positive ICP score.

Building the disqualification layer

Think of negative signals in three groups: identity signals, intent signals, and structural signals. Each one answers a different question about whether a signup is real, independent, and reachable.

Identity signals

These flag signups that aren't who — or what — they claim to be.

• Disposable or role-based email domains — mailinator.com, temp-mail.org, and similar throwaway providers should hard-fail. A fake signup detection layer usually catches these before enrichment even runs.
• Known competitor domains. If a competitor's engineering or product team signs up to see your onboarding flow, that's useful for your own research, not a sales lead. Keep a blocklist of competitor domains and route matches to a "competitive intel" tag instead of the sales queue.
• Education and nonprofit domains without a commercial use case. A .edu signup building a thesis project isn't disqualified forever, but it shouldn't inherit the same score as a company domain unless your product specifically targets education.
• Agency or freelancer domains evaluating on behalf of a client who never actually signs up. If the billing contact and the product user never match a real paying account within a defined window, flag it for review instead of letting it decay quietly in the pipeline.

Intent signals

These catch people who are technically real but have no path to becoming a customer.

• Usage that looks like one-off testing. A single session, no return visits, and no core action completed inside the trial window is a strong signal the account is inactive, not just slow to activate.
• Intent that points somewhere other than buying. Someone whose only activity is browsing your careers page and then creating a trial account is probably interviewing, not shopping.
• Geography you don't actually support. If your pricing, currency, or compliance posture doesn't cover a region, don't let a rep spend a cycle on a signup from outside it — route it to self-serve only.

Structural signals

These are about the account itself, independent of who signed up.

• Company size below your minimum viable deal size, if your product has a hard floor on seats or usage that makes a one-person company unprofitable to support with a human-touch motion.
• No active email domain or no detectable company at all. If enrichment comes back empty after multiple provider attempts in a waterfall enrichment setup, that's itself a data point. Treat "unknown" as its own bucket rather than letting it default into the regular queue.
• Already a customer under a different account. Duplicate signups from existing paying accounts should route to customer success, not new-business sales.

A simple disqualification checklist

Before a signup's ICP score reaches a rep, run it through a short gate:

1. Is the email domain disposable, role-based, or on the competitor blocklist?
2. Does company enrichment resolve to a real, identifiable organization?
3. Is the account inside a supported region and price range?
4. Is there any product usage beyond account creation?
5. Is this a duplicate of an existing customer account?

A "no" on question 2 or a "yes" on question 1 or 5 should remove the account from the sales queue outright. The rest feed into a weighted score alongside your positive ICP criteria, rather than gating completely — geography and low usage are softer signals that might just mean "not yet," not "never."

Where this plugs into your existing workflow

None of this replaces your ICP model — it sits in front of it. Groful runs enrichment and ICP scoring automatically on every signup, so the disqualification layer and the positive scoring model can run on the same enriched record instead of two disconnected systems.

Practically, that means:

• Enrichment resolves the company, domain, and any known duplicate-account matches.
• Disqualification rules run against that enriched data first.
• Anything that survives gets the full positive ICP score and routes to sales-assist or self-serve based on the result.

Teams using this layer tend to see the biggest change in the gray middle tier shrinking, not in the top-tier numbers moving. The best-fit accounts were always going to get attention. The win is that reps stop spending time on signups that were never going to close, and RevOps gets a cleaner view of which accounts in the CRM are genuinely worth forecasting against.

What to measure after rollout

Track these before and after adding disqualification rules:

• Rep hours per qualified meeting booked. This should drop as the gray middle tier shrinks.
• Percentage of CRM leads marked "unqualified" after a human touch. This should drop too, since those leads never reach a rep in the first place.
• False disqualification rate. Sample a slice of disqualified accounts monthly and check whether any should have gone through. This number should be low but nonzero — a rate of exactly zero usually means the rules aren't strict enough to matter.

If you're building this out and want to see how it fits your current enrichment and ICP setup, get in touch — most teams can wire the disqualification layer into an existing ICP model in a single pass rather than rebuilding scoring from scratch.