Lead scoring is a method for ranking individual prospects by how likely they are to buy, adding points for fit (such as job title and company size) and for behavior (such as visits, replies and demo requests), so sales contacts the most promising people first.
How lead scoring works
Lead scoring combines two kinds of evidence about a person.
- Explicit data is who they are: job title, seniority, company size, industry, country. It measures fit.
- Implicit data is what they have done: pages viewed, emails opened, webinar attended, pricing page visited, demo requested. It measures interest.
Each attribute and action earns points, some lose points (a student email address, a competitor's domain, an unsubscribe), and behavior points decay over time so a burst of activity six months ago stops counting. When a lead passes a threshold, it becomes a marketing qualified lead and goes to sales. People who have never raised a hand are a different problem, handled by cold outreach scoring.
The basic formula is a weighted sum: score = fit points + behavior points + negative points, with behavior points reduced by a decay factor based on age.
How to build a lead scoring model
- Start from closed deals. Look at the people involved in the last year of won deals. Which titles, company sizes and actions show up again and again?
- Assign fit points to the attributes that separated won from lost, not to every field you have. These should match your ideal customer profile.
- Assign behavior points by buying signal strength. A demo request is worth far more than a blog visit.
- Add negative points and decay. Subtract for clear non-buyers and reduce behavior points each month.
- Set the threshold with sales. Agree the score at which a lead is handed over. Seller time is what the score protects. McKinsey notes that traditional measures such as potential deal size might not help sellers use their time effectively, and that companies now use models to score leads and assign their best reps to the top prospects.
- Check monthly. Of the leads that crossed the line, how many became opportunities? Move points and the threshold accordingly.
Worked example: scoring two leads
Ledgerly, a fictional fintech selling payables software to manufacturers, uses a 100-point model with a handoff threshold of 60.
| Signal | Points | Lead A: finance director, 400 staff manufacturer | Lead B: analyst, 12 staff consultancy |
|---|---|---|---|
| Title is finance director or above | 20 | 20 | 0 |
| Manufacturer, 200 to 2,000 staff | 25 | 25 | 0 |
| Visited pricing page | 10 | 10 | 10 |
| Downloaded 4 guides | 3 each | 0 | 12 |
| Requested a demo | 30 | 0 | 30 |
| Not a target industry | minus 15 | 0 | minus 15 |
| Total | 55 | 37 |
Neither crosses 60. Lead A is a near-perfect fit who has not raised a hand yet, so marketing sends a targeted case study and an invite. Lead B is very active but a poor fit, and the negative points stop a demo request from a consultancy analyst landing on a rep's desk as if it were a buyer. Without the fit points and the penalty, B would have scored 52 and A only 10.
Lead scoring vs account scoring and predictive scoring
| Rules-based lead scoring | Predictive lead scoring | Account scoring | |
|---|---|---|---|
| Unit | One person | One person | One company |
| How points are set | By people, from judgment and closed deals | By a model trained on past conversions | Either, at the company level |
| Explainable | Yes, every point is visible | Depends on the model | Usually, if rules-based |
| Data needed | Little | Hundreds of past conversions or more | Firmographics plus contact activity |
| Best for | Early teams, simple sales | High volume inbound | Deals with several buyers |
Predictive scoring applies predictive analytics to the same question: models built to forecast new cases rather than explain old ones, a distinction Hofman and colleagues discuss in Integrating explanation and prediction in computational social science (Nature, 2021). It can find patterns people miss, but it needs enough history to learn from and it can be hard to explain to a rep who asks why a lead scored 87.
Common lead scoring mistakes
- Scoring activity, not intent. Email opens and blog visits pile up points from people who will never buy. Weight the actions that precede deals.
- No negative scoring or decay. Scores only go up, and the queue fills with stale leads.
- Ignoring the account. Three people from the same company each scoring 40 may be a stronger signal than one at 70. Roll up with account scoring.
- Setting it and leaving it. Without regular checks against outcomes, points stop matching reality. A RevOps scoring review cycle fixes that.
- Treating third-party signals as fact. Intent data from outside sources is probabilistic. Use it to prioritize, not to qualify on its own.
Lead scoring discipline applied to deals
ScoringFactory is not a lead scoring tool. Venture funds face a version of the same problem with inbound pitches (see inbound deal scoring), and the same discipline applies: decide what a good fit looks like, rank everything that arrives against it, and check the ranking against outcomes. ScoringFactory learns a fund's bar from the companies it backed and passed on, ranks founders and companies against it, and cites every score to the record. Investors also look at a company's lead scoring in diligence as part of its wider GTM scoring. To talk about your pipeline, contact the founders.
Frequently asked questions
What is lead scoring?
Lead scoring is a way to rank prospects by how likely they are to buy. Each lead earns points for fit, such as title and company size, and for behavior, such as visiting the pricing page or requesting a demo. Leads above an agreed threshold go to sales, so reps spend time on the most promising people first.
How do you build a lead scoring model?
Study who was involved in recent won and lost deals. Give fit points to the attributes that separated them and behavior points to the actions that came before a purchase. Add negative points for clear non-buyers and decay old activity. Set the handoff threshold with sales and check the conversion rate of qualified leads monthly.
What is the difference between lead scoring and account scoring?
Lead scoring rates one person. Account scoring rates the whole company, combining its fit with your ideal customer profile and the activity of everyone there. In sales with several decision makers, account scoring gives a truer picture, because buying interest is spread across a group rather than held by one contact.
What is a good lead scoring threshold?
There is no standard number. The right threshold is the score above which leads convert to opportunities at a rate sales finds worth its time. Start with a guess, track conversion by score band for a month or two, and move the line to where conversion clearly rises. Revisit it whenever the model changes.