How lead scoring works
A score is built from two kinds of evidence, and keeping them apart is what makes it useful. Fit is who the person is: their industry, the size of the business, the country, the role they hold, the budget bracket implied by what they asked for. Behaviour is what they did: pages read, a pricing page opened more than once, a form completed, an email clicked, a call booked.
Each signal adds or removes weight, and the running total sits on the contact record where sales can see it. Subtraction matters as much as addition. A country you do not serve, a role with no authority to buy, a free address on an enquiry about enterprise work — those should pull a score down rather than simply failing to lift it, otherwise the model rewards activity instead of suitability.
Why lead scoring matters
Once enquiries arrive faster than everyone can be called properly, somebody is already choosing an order. Without a score that choice falls to whoever shouted loudest or wrote most recently. A score makes the order visible and arguable, which is the point: a rule you can see is a rule you can correct.
It also gives marketing and sales one vocabulary. When both sides agree what a strong lead looks like, the argument moves from “the leads are bad” to “this signal is weighted wrongly”, which is a question evidence can settle. And it makes automation safe, because a follow-up sequence that fires on a score behaves differently for a serious buyer than for a curious student.
Common mistakes with lead scoring
The first is inventing the weights. A model assembled in a meeting reflects what the room believes, not what the business wins, and it will confidently promote the wrong people for months. Build it backwards from customers you actually closed, and check the traits they truly shared.
The second is scoring behaviour only. Someone who reads every blog post and opens every email may be a competitor, a job seeker or a student. Without fit signals holding the score down, they outrank a buyer who arrived once, read the pricing page and enquired. The third is letting scores accumulate forever, so an old lead who has gone cold keeps a total earned a year ago. Interest fades, and the score should fade with it.
How to act on it
Start with your ideal customer profile and give real weight to the handful of traits your won customers share. Keep the model small enough to explain in a sentence; a model nobody can explain is one nobody will trust or maintain. Agree the threshold at which sales must call, and write down what happens below it.
Then close the loop. Every month, look at which scores actually converted and adjust the weights that got it wrong. Feed the result into lead routing so high scores reach a person quickly, and use what you learn to change where the lead generation budget goes — scoring is a measurement of your sources as much as of individual people.