Lead scoring sounds technical, but a useful model fits on one page. This guide explains the two scores every model needs, how to build one from your own deals, a worked scoring sheet you can copy, how to set the threshold for sales, and how to tell whether the model works.
01
What a lead scoring model is
Lead scoring ranks leads by adding points for things that make a deal more likely and taking points away for things that make it less likely. The model is the list of those things and their points, plus the rule for what happens at each score.
It answers one question: which leads should a person look at first? It does not decide whether a lead is worth pursuing. That still takes a conversation, as covered in how to qualify leads.
02
Fit and activity, scored separately
| Score | What it measures | Typical inputs |
|---|---|---|
| Fit | Whether the company and person look like your customers | Industry, size, region, systems they run, the person's role |
| Activity | Whether they are showing interest now | Demo requests, pricing page visits, several people from one company, replies |
Keep the two apart. A perfect-fit company with no activity is an outbound target, not a hot lead. A very active lead that does not fit is a student, a competitor or a researcher. One combined number hides both cases.
03
How to build a lead scoring model, step by step
- Pull your won and lost deals. The last year is usually enough. Include the leads that never became deals, too.
- Find what separates them. For each attribute, count how often it appears in won deals and in the rest. Write it as a count with the base, such as "18 of 25 won deals were distributors".
- Give points to the differences. Attributes common in wins and rare elsewhere get the most points. Attributes that are equally common in both get none.
- Add activity points, and let them fade. A pricing page visit last week matters; one from six months ago does not.
- Subtract for clear mismatches. Competitors, personal email addresses, regions you do not serve.
- Set the routing rule. Decide which combination of scores goes to sales, which goes to nurture, and which gets no sales time.
- Test it on the past. Score last year's leads with the new model and check that the winners score higher.
04
A worked lead scoring sheet
The sheet below is an invented example for a company that sells software to wholesale distributors. The points are made up to show the shape; yours should come from your own deals.
Scoring sheet · invented example
LEAD SCORING SHEET (invented example: software for wholesale distributors) FIT SCORE (who the company is) +20 Industry is wholesale distribution +15 50 to 500 staff +10 Operates in our sales regions +10 Runs a system we integrate with -50 Is a competitor or a reseller -20 Personal or student email address ACTIVITY SCORE (what they have done, last 30 days) +30 Requested a demo +15 Visited the pricing page +10 Two or more people from the same company +5 Downloaded a guide +1 Opened a newsletter Halve the activity score after 30 days with no activity ROUTING To sales within one working day: fit 40 or more AND activity 30 or more Outbound or nurture: fit 40 or more, activity under 30 No sales time: fit under 40
In the example, a 200 person distributor in your region that requested a demo scores 45 on fit and 30 on activity, so it goes to sales. A 3,000 person retailer whose student intern downloaded two guides scores below zero on fit, so it gets no sales time, however active it looks.
05
How to set the threshold for sales
Start from capacity, not from a round number. If your team can follow up a certain number of leads properly each week, set the threshold so roughly that many cross it. A threshold that sends sales more leads than they can call within a day or two turns scoring into noise.
Review it often in the first weeks. If sales keeps rejecting leads that crossed the line, raise it or change the inputs. If good deals keep arriving from below the line, lower it.
06
How to test whether the model works
Every quarter, take the leads scored in that period and compare them with what happened. Leads that scored high should have become deals more often than leads that scored low. If they did not, the model is ranking on the wrong things.
Look at the misses by hand. When a deal closes from a low score, look for an input the model lacks, such as a dated event at the company: a new site or a new leader. Those events are buying signals, and they are worth adding as their own points.
07
Common lead scoring mistakes
Points from opinion. A model built in a meeting reflects what people believe, not what closed.
One combined score. Fit and activity mixed together hide the leads that need different treatment.
Rewarding clicks. Email opens and newsletter clicks are cheap signals and easy to inflate.
Activity that never expires. A lead stays hot for months because of a visit nobody remembers.
Never testing it. A model that is not checked against closed deals drifts away from reality.
Treating the score as the decision. The score orders the queue; a conversation decides.
08
The lead scoring checklist
Before you switch it on
- Points come from your won and lost deals.
- Fit and activity are scored separately.
- Activity points fade with time.
- Clear mismatches subtract points.
- The threshold matches what sales can follow up.
- A date is set to test it against closed deals.
09
Where to go next
Fit scores are only as good as the profile behind them. Write yours with this ideal customer profile template, and find the questions that confirm a high score in these discovery call questions.