ICP scoring: the profile-fit dimension done right
ICP scoring sits at the heart of every B2B lead scoring model, and it’s the dimension most teams treat as static. Your ideal customer profile drifts continuously. Every product launch, every move upmarket, every pricing change reshapes who you actually sell to. The scoring rules don’t drift with it unless someone makes them.
This post is the second in a series of deep dives on the four dimensions of a working lead scoring framework. The first covered engagement, the activity a lead shows right now. This one covers fit: the person and the company you sell to. There’s a strong case for splitting those into two separate dimensions, profile fit and account fit, rather than treating them as one. We’ll start with why, then walk through how to build a profile-fit score that holds up.
Profile fit vs account fit: why the split matters
Most lead scoring tools, including HubSpot’s modern lead scoring tool, expose “Fit” as a single score. Lumping profile fit and account fit into one number hides which side of the model is failing when scores stop predicting.
Profile fit scores the human attached to the contact: title, seniority, function, decision-making authority, geographic location, and any signals about budget or timeline.
Account fit scores the company: industry, size, tech stack, growth signals, geography, and vertical match.
These behave differently as your business changes:
- Move upmarket on pricing, and account fit shifts (target company size grows) before profile fit does (the buyer persona stays similar).
- Launch a new product for a different buyer, and profile fit shifts (new persona) while account fit might not (same target accounts).
- Expand into a new region, and both shift, but at different rates and through different signals.
Collapsing them into one number hides which one is drifting. We recommend scoring them as separate dimensions even when your tooling joins them into a single “Fit” field.
This post covers profile fit specifically. Account fit gets its own post.
The signals that matter for profile fit
The signals worth scoring, ranked by predictive value:
Title and seniority
- C-level or VP at target accounts: heavy positive
- Director / Head of [function]: positive
- Manager / Senior Manager: small positive
- Individual contributor in target function: neutral or small positive
- IC outside target function: zero
- Junior titles (Associate, Analyst): zero
- Disqualifying titles (Student, Intern, Unpaid Volunteer): negative
The strongest title signals are the role and seniority combinations that match your buyer persona. “VP of RevOps” and “Director of Marketing Operations” should both score high. “VP of Engineering” probably shouldn’t, unless you sell to engineering buyers.
Function
Function is harder to capture than title because most companies don’t expose it as a clean property. You’re either inferring from title (which works about 80% of the time) or asking on forms (“What’s your role in evaluating tools like ours?”).
Score positive on whichever functions actually buy your product:
- For RevOps tools: RevOps, Sales Operations, Marketing Operations, Demand Gen
- For developer tools: Engineering, Platform, DevOps, SRE
- For finance tools: Finance, FP&A, Accounting, Treasury
Decision authority signals
Where a lead gives you explicit signals, score them positive:
- Mentions a budget number on a form
- Mentions a specific timeline (“evaluating in Q3”)
- Mentions ongoing vendor evaluation (“comparing 3 tools”)
- Sales notes indicate budget authority
These signals are gold when you can capture them, but they’re rare. Don’t penalize leads who don’t provide them.
Email domain credibility
A real corporate domain is a small positive signal. A free-tier address (gmail, yahoo, personal outlook) on a tool that targets corporate buyers is a small negative.
Negative profile fit signals
Most teams underuse negative scoring on profile fit. The signals worth deducting on:
- Job title is “Student,” “Intern,” or similar (−10 to −15)
- Job title is “Founder” at a one-person LLC, on an enterprise product (−5)
- Free-tier email domain on a corporate product (−5 to −10)
- Title contains terms unrelated to your buyer persona (−5)
Negative scoring on profile fit keeps the model from drifting upward forever. A lead who downloads three whitepapers and holds a junior title should not score “Hot” on activity volume alone. This is the same reason engagement needs negative scoring and decay: without a way to move down, every score eventually climbs into the red zone and stops meaning anything.
Where profile fit drifts
Profile fit is the dimension that changes most with positioning shifts. The drift signals:
You changed positioning. New homepage, new messaging, new buyer persona, and scoring rules still aimed at the old persona. Refresh them.
You launched a new product. The new product has a different buyer, often a different function or seniority. Build separate profile-fit rules per product line if you sell to materially different personas.
Your sales motion shifted. You used to run PLG with a self-serve buyer; now you’re running top-down enterprise sales. The buyer changes from end-user to executive sponsor, and profile-fit rules need to follow.
You expanded into a new region. Titles differ across markets. The “Director of Operations” in the US is a “Head of Ops” in the UK and a different role entirely in Japan. Adjust the rules per region.
The fix for any of these is the same: rewrite the profile-fit rules against your most recent closed-won deals. Look at who actually bought, not who you wish would buy. Stale fit rules are one of the quiet reasons HubSpot lead scoring stops working a couple of quarters after go-live.
How to validate
The validation loop:
- Pull every closed-won deal from the last 90 days.
- Look at the primary contact (or buying committee) on each.
- Score each contact retrospectively on profile fit using your current rules.
- Compute the average profile-fit score for closed-won leads.
- Compare it to the average profile-fit score across all your MQLs.
If closed-won contacts score higher than the average MQL on profile fit, your rules are doing useful work. If they score the same or lower, your rules are inverted, rewarding the wrong personas, and need rebuilding.
Run this quarterly. Profile fit is the dimension where drift hits fastest after a business change.
Common mistakes
A few traps:
Scoring “knows about us” as profile fit. A lead who has visited your homepage three times is engaged, not a fit. Keep engagement signals out of the fit dimension.
Inferring function purely from title. “Marketing Director” could mean Demand Gen, Brand, Field Marketing, Product Marketing, or Comms. If function matters to your scoring, ask explicitly on forms or use enriched data.
Static rules forever. The most common failure mode. Schedule a quarterly review.
Treating profile fit as a checklist. Real ICP fit is fuzzy. A senior PM at a target account isn’t your buyer on paper, but they might be the influencer who pulls the buyer in. Soft positive scoring for adjacent personas, not zero and not full points, handles this better than binary rules.
Where this fits
Profile fit is one of four dimensions in a complete lead scoring framework. The other three are account fit (the company), engagement (what they’re doing now), and deal context (open opportunities). Profile fit alone is a partial picture. A perfect-fit person at a wrong-fit account is still not a buyer, and a high-fit contact with zero engagement is a name on a list, not an MQL.
A note on tooling
If you want a scoring system that exposes profile fit as a separate dimension from account fit, with calibration tooling that compares your fit rules to actual closed-won data on demand, that’s what kenbun ships. See it on your HubSpot data.