A lead score nobody can argue with is a lead score nobody uses

Salespeople reject scores they cannot check

Give a sales team a list of leads scored 0 to 100 and watch what happens. They work the top of the list for a week, find two that are obviously wrong, and go back to working the list in the order it arrived.

This is a rational response. A score with no visible reasoning is an assertion, and the first time it is wrong there is no way to tell whether it was unlucky or systematically broken.

Score, tier, reason — in that order of importance

The reason is what makes the system usable. It should be checkable against the submission in about five seconds:

Hot, 88. Named event with a date in eleven weeks, 1,800 delegates stated, asked specifically about badge printing throughput. Title is Head of Events. Next: send the registration throughput calculator and offer a scoping call this week.

A salesperson reads that, glances at the enquiry, and either agrees or does not. When they disagree they can say why, and that is how the scoring gets better.

Compare it with "88". There is nothing to agree or disagree with.

What a score should be built from

Signals that are actually in the submission:

• //A named event with a date. The single strongest signal, because it implies a budget cycle already in motion.
• //A stated delegate count or scale. Turns an enquiry into something quotable.
• //A specific service question. "How fast is your check-in" is a buyer. "What do you do" is research.
• //A decision-making title, where one is given.
• //An existing supplier mentioned. Often the strongest signal of all — they have a budget and a problem with their incumbent.

What a score must never be built from

Email domain as a proxy for seniority. A director using a personal address during a switch is not a weak lead, and treating them as one is how you lose the deal you most wanted.

Company name as a proxy for budget. A large organisation's brand-marketing team may have less budget than a mid-size firm's annual conference.

Message length. Short enquiries from busy people are common. Long enquiries from students are also common.

These correlate weakly in aggregate and produce confidently wrong individual judgements, which is precisely the failure that destroys trust.

The tiers that matter operationally

Five tiers, because three is not enough to separate "act today" from "act this week":

• //Hot — a concrete opportunity to act on today
• //Warm — a real requirement missing timing or scale
• //Cool — genuine but exploratory
• //Unqualified — out of scope, including job applications and vendor pitches
• //Spam — link injection and bulk marketing

The last two exist so the first three stay clean. A sales list containing job applications gets abandoned.

Score at write time, never in the request path

The scoring call belongs after the lead row is saved, not before it. If the model is slow, or the API key has expired, or the budget cap has been hit, the enquiry must still land.

An unscored lead is an inconvenience. A lost lead is revenue, and it is invisible — nobody knows to look for the enquiry that never arrived.