Fit Score
A fit score measures how well a lead matches your ideal customer profile based on firmographic and demographic attributes.
Key takeaways
- Express fit as a percentage of available weight, not a raw sum.
- Build the profile from customers who renewed, not from everyone who signed.
- Fit answers suitability; it says nothing about whether they are ready.
- A backward-looking model undervalues any segment you have just entered.
- One quiz result can serve both internal routing and respondent feedback.
In depth
A fit score compares one lead against a profile built from customers who succeeded. Each attribute in the profile, industry, headcount band, region, role, existing tooling, gets a weight reflecting how strongly it separates good customers from the rest, and the lead's score is the share of available weight it matches. Expressing it as a percentage of the maximum rather than a raw sum means the number keeps its meaning when attributes are added or removed.
A fit score is only as good as the definition of a good customer behind it. Building the profile from everyone who ever signed produces a wide, useless target; building it from customers who renewed and expanded produces a narrow, demanding one that rejects deals your sales team could still win. The number of attributes matters too: too few and everyone scores similarly, too many and the score becomes a description of your existing customers rather than a filter for new ones.
Fit is normally the first gate in the funnel and the one that decides how much a lead is worth spending on. High-fit leads justify a call, a tailored proposal and patience; low-fit leads get self-serve material. In a scorecard funnel the fit score is computed at submission and shown back to the respondent as a tier, which means the same number simultaneously serves an internal routing decision and an external piece of feedback the respondent came for.
Fit describes suitability, never readiness. A perfect-fit company with no trigger event will not buy this year regardless of its score, which is why a high fit score alone should never dictate urgency. The model is also backward-looking by construction: it encodes who bought before, so it will systematically undervalue a new segment you have just started to serve. Treat a persistently low-scoring group that keeps converting as evidence to recalibrate, not as noise to be ignored.
Example in practice
How to measure it
The headline test is lift: win rate among high-fit leads divided by win rate across all leads. A ratio near one means the score is not separating anything. Read it alongside average deal size and retention by fit band, because a fit score that predicts closing but not renewal is selecting for people who say yes, not for customers who stay.
Watch the distribution as well as the outcomes. A healthy fit score spreads leads across its range; one that piles everyone into the middle is measuring attributes that barely vary in your traffic. Track the share of closed deals that scored below your sales threshold, since every one of them is a lead the model told you to ignore and a direct argument for changing a weight.
Common mistakes
The classic error is describing your customers instead of predicting new ones. If most current customers are agencies, agency scores highest, and the model recommends more of what you already have while missing an adjacent market that converts just as well. Include losses in the calibration, not only wins, and check whether each attribute separates customers from leads who were qualified and still did not buy.
The second is leaving the score frozen after launch. Fit scoring is usually configured during a positioning exercise and then never touched, so it keeps scoring against a market you left two product versions ago. Put a recalibration date in the calendar, and treat any change in pricing, packaging or target segment as an automatic trigger to review the attribute list and its weights.
Frequently asked questions
How often should I update my fit score model?
Revisit it at least quarterly or whenever your ICP shifts, such as moving upmarket or entering a new vertical. A stale model keeps prioritizing customers who no longer reflect your best business.
What attributes belong in a fit score?
Only ones you can observe for most leads and that visibly differ between customers and non-customers. Industry, company size, region and role are the usual starting set. Add an attribute about the buyer's situation, such as whether the process is currently manual, since that often separates better than firmographics. Drop anything you can only fill in for a minority of records.
What is the difference between a fit score and a lead score?
Lead score is the umbrella term and often mixes fit with behaviour into one number. Fit score is strictly the ICP-match half: it changes only when an attribute changes, not when someone visits a page. Keeping fit separate means you can answer whether the pipeline problem is the wrong audience or the right audience at the wrong moment.
How do I build a fit score without much data?
Start with judgement and make it explicit. List the ten best customers, write down what they had in common before they bought, and turn those commonalities into three or four weighted attributes. It will be rough, but a written model can be tested and corrected, whereas an unwritten sense of who is a good lead cannot be argued with or improved.
Should a low fit score disqualify a lead automatically?
Not on its own, unless a hard constraint is involved such as an unsupported region or a regulated industry you cannot serve. A low score is best treated as a decision about effort: no outbound call, no custom proposal, but still access to self-serve options. Automatic deletion removes the chance to notice that the model is wrong.
Can a fit score predict churn?
It can, if it was built from customers who stayed rather than customers who signed. When retention is part of the definition, a low-fit customer who bought anyway is a predictable renewal risk, and that is worth flagging at handover to onboarding. If the profile was built purely from closed-won deals, it predicts buying and says nothing about staying.
How many fit tiers should a scorecard produce?
Usually three, because each tier needs its own result page copy and its own follow-up. Three lets you say clearly that a respondent is a strong match, a partial match, or better served elsewhere. A fourth tier is worth adding only when a genuinely different offer exists for it, such as a partner referral for companies you cannot serve directly.