Lead Fit Score
A lead fit score is a numeric grade that measures how closely a prospect matches your ideal customer profile based on firmographic and demographic attributes.
Key takeaways
- A fit score is arithmetic: weighted points summed and normalized to a fixed range.
- How blank answers are handled changes the ranking more than the weights do.
- Hard requirements belong as gates outside the score, not as negative points.
- Store the top contributing attributes beside the total so reps can read it.
- Version the model, because scores from different weightings are not comparable.
In depth
The score is an arithmetic object with a defined range. Each attribute contributes points, either a fixed value per answer or a share of a category maximum, and the sum is normalized so that a band means the same thing across every form and quiz feeding it. Two design choices govern its behavior more than the weights do: whether an attribute may contribute negative points, and whether a missing answer scores zero or is removed from the denominator entirely.
Weights should reflect separation, meaning how differently an attribute appears among won and lost accounts, but they also have to stay explainable to the people acting on them. A model with twenty finely tuned inputs may separate marginally better than one with six and be trusted far less. More attributes also mean more blanks, and each blank forces the normalization question again. Simplicity buys adoption; complexity buys accuracy nobody uses once reps stop believing the number.
A score is only useful next to its reason. Store the total, the band, and the two or three attributes that contributed most, so a rep opening the record sees why it scored as it did rather than a bare figure. In a scorecard funnel the arithmetic runs during the session itself: each answer carries points, categories roll up into a total, and rating tiers convert that total into a named band that drives both the result page and the routing.
The number compresses a handful of answers, and compression loses the exception. An account failing one hard requirement can still score in the seventies because everything else is strong, which is why blocking conditions belong outside the arithmetic as gates rather than inside it as negative points. Scores also become incomparable after a re-weighting, since a seventy from last quarter no longer means what a seventy means today unless the model version is stored with it.
Example in practice
How to measure it
Look at the distribution first. A healthy score spreads leads across bands rather than piling them at one end, and a heavy pile usually traces back to a single attribute almost everyone answers the same way. Then check that opportunity rate rises from band to band; any band converting worse than the one below it points to an input carrying the wrong weight.
Also measure stability and agreement. Re-score a sample of last year's customers and check whether your best accounts would clear today's threshold, because if many would not, the weights have drifted away from reality. Compare band assignments against how reps rank the same leads by hand, since large disagreements usually locate a missing attribute rather than a stubborn rep.
Common mistakes
The most common mistake is awarding points for answering rather than for the answer. When every option in a dropdown scores something, a lead's total rises simply by completing the form, and the most diligent respondents outrank the best-fitting ones. Give zero to options that indicate no fit, reserve points for answers that genuinely match the profile, and inspect the resulting distribution for an artificial floor that no lead ever falls below.
The second is inflating the scale to look precise. Scoring out of a thousand implies a resolution six form questions cannot support, and reps begin arguing over ten-point gaps that carry no meaning. Keep the range modest, define three or four bands each tied to a specific action, and treat any two scores inside the same band as equivalent when deciding who gets called first.
Frequently asked questions
What is the difference between fit score and lead score?
Lead score is often an umbrella term that blends fit and engagement, while fit score isolates only how well the prospect matches your ideal customer profile. Keeping them separate lets you tell an interested-but-wrong-fit lead apart from a perfect-fit lead who is not yet engaged.
What data do I need to calculate a lead fit score?
You need firmographic and demographic attributes such as company size, industry, role, region, and use-case match. In a quiz funnel these are gathered directly from the questions a respondent answers, removing the need for third-party enrichment in many cases.
How often should I recalibrate my fit scoring model?
Review the weights every quarter or after any major shift in your ideal customer profile or pricing. Compare which scores correlated with closed-won deals and adjust the point values for attributes that proved more or less predictive.
What scale should a lead fit score use?
Zero to one hundred is the usual choice, because bands are easy to communicate and the numbers survive re-weighting reasonably well. What matters more than the range is the number of bands: three or four, each tied to a specific action. A finer scale implies a precision that a handful of form answers simply cannot deliver.
Should a lead fit score include negative points?
Use them sparingly, and only for attributes that genuinely reduce fit, such as a competing product already embedded across the account. For anything that makes a deal impossible, a gate works better: the record is marked disqualified regardless of the total, so strong scores elsewhere cannot mask a blocking condition that ends the deal anyway.
How do I score a lead with missing attributes?
Decide in advance whether a blank scores zero or is removed from the denominator, then apply that rule consistently. Scoring blanks as zero punishes incomplete forms and pushes good leads down; excluding them inflates scores built on very little data. A workable middle path is to score what is present but flag records below a minimum answered-attribute count.
How many attributes should feed a fit score?
Usually between four and eight. Below four the total swings on a single answer; above eight each additional attribute adds little separation while raising the blank rate and making the number harder to explain. Add an attribute only when you can show that it separates won from lost accounts on its own, not merely that it is available.
What is the difference between a fit score and a fit tier?
The score is the raw total; the tier is the band it falls into together with the label attached to that band. Teams act on tiers because a label maps to one agreed action, while raw numbers invite negotiation. Keep the score for analysis and threshold tuning, but drive routing and reporting from the tier itself.
How do I explain a low fit score to a sales rep?
Show the contributing attributes rather than the total. A rep accepts that an account scored low because it has five employees and no relevant tooling far more readily than a bare figure, and the stated reason lets them challenge the model when it is wrong. Record those reasons automatically so nobody has to reconstruct them weeks later.