What is lead scoring
Also known as: lead score, lead scoring model
Lead scoring is the method of assigning points to each lead to rank who deserves attention first. The score combines fit data, such as company size, industry and job title, with behavioral data, such as website visits, downloads and email replies. When a lead passes a threshold, it goes to sales; below it, it keeps receiving content.
Lead score = fit points + behavior points − negative points
The threshold for sending a lead to sales should come from leads that already became customers, not from guesswork. Revisit it when the close rate of leads above the threshold drops.
A software company gets 400 leads a month and its sales team can only call 80. Before scoring, the order was first come, first served, and the best contacts waited days.
With a simple points table and a threshold of 60, about 70 leads a month go to sales, precisely those with the right profile and real interest. The team calls them first, and the rest get an email sequence until they show more signals.
Fit and behavior
Scoring models combine two kinds of data. Explicit data is provided by or about the lead: company size, industry, job title, location. Implicit data comes from behavior: website visits, downloads, email opens and clicks.
Taking points away also matters: a personal email address for a business product, a student, a competitor, a temporary email address. Negative points keep click volume from hiding a lead that will never buy.
Example points table
| Points | |
|---|---|
| Company in the ideal customer profile's industry | +20 |
| Between 50 and 500 employees | +15 |
| Decision making role | +15 |
| Visited the pricing page | +20 |
| Requested a demo | +40 |
| Opened three or more emails this month | +5 |
| Personal email for a business product | −10 |
| Student or competitor | −50 |
Rules, prediction and AI
The most common model is rule based: someone defines criteria and points, as in the table above. A predictive model uses machine learning on customer history, sometimes combined with third party data, to find on its own which signals come before a sale.
Predictive scoring needs volume: with a few dozen sales a year, well thought out rules usually work better than a model without enough data to learn from.
How to build it in five steps
- Start with customers who closed: their profile and what they did before buying.
- List the signals you can measure, on the website, in email and in the CRM.
- Assign weights, with more weight for intent signals such as demo requests.
- Set the threshold together with sales.
- Every quarter, compare scores with actual closes and adjust.
Common mistakes
- Weighting what is easy to measure, such as email opens, over what matters.
- Never decaying points over time; a lead who viewed pricing a year ago is not yesterday's lead.
- Building the table without sales, who then do not trust the score.
- Never revisiting the model.
Frequently asked questions
What is lead scoring?
The method of assigning points to each lead, combining fit and behavior, to rank who sales should contact first.
What data goes into lead scoring?
Fit data, such as company size, industry and job title, and behavioral data, such as website visits, downloads, email opens and demo requests.
What is the difference between lead scoring and an MQL?
Lead scoring is the score. An MQL, a marketing qualified lead, is usually a lead that passed the threshold defined in that score.
What is predictive lead scoring?
Scoring done by a machine learning model that analyzes customer history to find which signals come before a sale.
How often should a scoring model be reviewed?
Quarterly is a good pace: compare lead scores with what actually closed and adjust weights and the threshold.
Related terms
- MQL and SQLAn MQL (marketing qualified lead) is a lead that marketing considers ready for a sales approach, because it...
- ICPAn ICP, short for ideal customer profile, is the kind of company, or customer, that gets the most out of what...
- Lead nurturingLead nurturing is a planned relationship with contacts who are not yet ready to buy: a sequence of messages...
- Marketing automationMarketing automation is the use of software to run repetitive marketing tasks and bring together in one...
- CRMCRM, short for customer relationship management, is the process and, almost always, the system that records...
- LeadA lead, in marketing, is a person or company that showed interest in what you sell and left a way to be...
Rodrigo Fávaro
Founder of ROO3, a marketing and technology agency in São José do Rio Preto, Brazil. Builds AI products running in production (Tobia, gerar.app, Pense Mercado) and maintains the AI Benchmark, a public ranking of AI models.
X @rodmf LinkedIn rodrigofavaroMarketing you can measure, from the ad to the sale.
ROO3 builds and runs paid media, SEO and websites with real targets, dashboards and numbers. The first conversation is free.