Lead Scoring

Lead scoring ranks leads by how likely they are to buy, so a sales team with more leads than hours works them in the right order. It answers one question: of everyone who has raised a hand, who should be called first?

Two kinds of signal

Every workable model combines two things, and a model using only one of them fails in a predictable way.

Fit — who they are. Company size, industry, role, region, whether they match the customers you already serve well. Fit is knowable before the lead does anything at all.

Behaviour — what they have done. Pages viewed, pricing page visits, demo requests, email replies, documents downloaded, return visits.

A model scoring only fit promotes a perfect-profile company with no current intent. A model scoring only behaviour promotes a student doing research and a competitor reading your pricing page. Both are common; both waste the same sales hours.

A worked model

Signal Type Points
Company size 50–500 staff Fit +15
Industry is banking, insurance or telecom Fit +20
Job title contains "head", "director" or "chief" Fit +15
Personal email domain (gmail, yahoo) Fit −20
Visited the pricing page Behaviour +25
Requested a demo Behaviour +30
Returned within 7 days Behaviour +10
No activity in 30 days Behaviour −15

A banking operations director from a 200-person firm who viewed pricing and booked a demo scores 105. A personal-email visitor who read one blog post scores −20. The ordering is now obvious and does not need a meeting to agree.

Setting the threshold

The threshold is where a lead becomes sales-ready — commonly called a marketing qualified lead, or MQL.

Set it from your own closed-won history rather than by choosing a round number. Score your last hundred closed deals as they were at the moment sales first engaged, and find the score at which win rate rises sharply. That is your threshold.

A threshold set too low floods sales with leads that do not convert, and the team stops trusting the score. Set too high, leads sit while a competitor calls them. Either way the failure shows up as sales ignoring the score entirely.

Score decay

Intent is perishable. Someone who requested a demo four months ago and has done nothing since is not the same lead as someone who requested one yesterday, even though both have the same event in their history.

Behaviour points should decay — a common approach reduces them by half after 30 days of inactivity. Fit points should not decay: the company is still the same size and in the same industry.

An unexplainable score gets ignored

This is the practical failure of most lead scoring, and it is not a modelling problem.

When a rep sees a lead scored 87 and cannot see why, they fall back on their own judgement, and the score becomes an ornament. A score that shows its working — these signals, these weights, this total — is a score a rep will argue with, correct, and then use. The arguing is the point: it is how the weights get fixed.

For the same reason, the weights have to be yours to change. A model you cannot tune is a model that stays wrong about your market after the first time it is wrong.

Avant One℠ CRM scores leads on twelve signals with the weighting set by you and the reasoning shown on the record, so a rep can see which signals produced a score before deciding whether to trust it.

Frequently asked questions

What is the difference between lead scoring and lead qualification?

Scoring is automatic and continuous, ranking every lead as signals arrive. Qualification is a human judgement at a point in time, usually a conversation confirming budget, authority, need and timing.

What is an MQL?

A marketing qualified lead: a lead whose score has crossed the threshold at which it is handed to sales. The threshold should come from your own closed-deal history.

Should lead scores expire?

Behaviour points should decay with inactivity, commonly by half after about 30 days. Fit points describe the company rather than its intent and should not decay.

Why do sales teams ignore lead scores?

Almost always because the score cannot be explained. A number with no visible reasoning is overridden by the rep's own judgement, and the model never gets corrected because nobody can see what to correct.

Turn customer experience into an operational advantage

See how Avant One unifies support, outreach and collections with AI built into every workflow.