Lead scoring gives each lead a number that estimates how likely it is to become a customer. Sales and marketing teams use the score to decide who to contact first, who needs more nurturing and who is unlikely to buy at all.
How lead scoring works
There are two main approaches.
Rule-based scoring
You decide which signals matter and how many points each is worth. Signals fall into two groups. Fit describes who the lead is: job role, company size, sector, location within your service area. Behaviour describes what they have done: requested a quote, viewed the pricing page, attended a webinar, replied to an email. Negative points cover signals that suggest a poor fit, such as a competitor’s domain or a location you cannot serve. Most systems also let behaviour points fade over time, so a burst of interest last year does not keep someone at the top of the list.
As an illustration of the structure, not a benchmark, a B2B software firm might award points for a demo request, fewer for a pricing page visit, a few for a matching job title, and deduct points for a personal email address on a business product. When a lead passes an agreed threshold, they become a marketing qualified lead and are passed to sales.
Predictive scoring
Instead of you choosing the points, a machine learning model studies past leads in your CRM, compares those that became customers with those that did not, and scores new leads on how closely they resemble the winners. It needs a reasonable history of won and lost deals to be reliable, so it suits businesses with steady lead volumes more than those closing a handful of deals a year.
Why it matters
Scoring focuses limited sales time where it is most likely to pay off, and it gives marketing a shared definition of a good lead instead of an argument. It also improves paid advertising: if you send qualified leads back to Google Ads, for example through enhanced conversions for leads, the bidding systems can learn to find more people like them rather than more form-fillers.
In the UK, lead scoring counts as profiling under UK GDPR. You need a lawful basis, which for B2B scoring is usually legitimate interests backed by a written assessment, and you must explain the scoring in your privacy notice. People have an absolute right to object to profiling used for direct marketing. If a score were ever used to make a decision with a significant effect on someone, such as refusing them credit, stricter rules apply.
Common mistakes
- Awarding points for email opens. Apple Mail Privacy Protection registers opens automatically, which inflates scores for anyone using Apple Mail.
- Choosing points by instinct and never checking them against actual sales.
- Building dozens of rules that nobody can explain.
- No decay, so old activity keeps leads looking hot.
- Setting the threshold without sales agreeing to it, so they ignore the scores.
How to act on it
- Export the last twelve months of won and lost leads and look for what the winners had in common.
- Build a simple model with five to eight signals, mixing fit and behaviour.
- Agree the hand-over threshold with whoever does the selling.
- Each quarter, compare close rates across score bands. If high scores do not close better than low ones, change the model.
- Update your privacy notice and record your lawful basis.
Linking lead quality back to channel spend is central to how I run performance marketing.