Lead scoring explained: models, a worked B2B example and the mistakes to avoid

Lead scoring gives each prospect a number so you know who to contact first. It sounds simple, and the arithmetic is. The hard part is choosing what to score, and being honest about what the number means.

The three common models

For outbound, most small teams should start with points-based fit plus intent: rules you can read, split between "is this the right company and person?" and "is there a reason now?".

A worked 100-point example

Say you sell prospecting software to small B2B software companies, and your best buyers are founders or sales leaders at 11 to 200-person companies that just posted a sales role.

A founder at a 40-person software company in the US who posted an SDR role two weeks ago, started as CEO three years ago and has a verified email scores 25 + 20 + 10 + 10 + 0 + 5 + 5 + 5 + 0 + 5 = 85. A marketing manager at the same company scores far lower, which is right: same company, wrong person.

The weights above are a reasonable start, not a law. After a month, check which factors your replies actually had in common, and move points toward those.

Rules that keep a score honest

• Show the breakdown. A number nobody can explain gets ignored. Every score should come with the lines that add up to it.
• Unknown is not zero. If you don't know a company's size, say "unknown" instead of scoring it as a mismatch, and make the difference visible.
• Only score what the ICP asks about. If location doesn't matter to you, leave it out instead of giving everyone the points.
• Company signals only count for the right person. A hiring company's engineer shouldn't get the hiring points meant for the sales leader.
• It's priority, not probability. An 80 means "contact before a 60", not "80% likely to buy". Say so.
• Change the version when you change the rules, so old and new scores aren't compared as if they meant the same thing.

A mistake we made: ties

Our first live run of the Audyense score gave 20 prospects exactly the same score, 55. Every one of them matched the same search filters, so every fit factor was identical. That's why "new in role" exists: it's a real difference between people who otherwise look the same, because people new in a role make more decisions. If your scores bunch up, you're missing a factor that actually separates people, and adding a hidden tie-breaker only hides that.

Using the score

• Contact the top of the list first, but read each one; a score is a sorting tool, not a decision.
• Mark outcomes (replied, meeting, not a fit, and why) so you can tune the weights.
• Recheck weights monthly with at least 20 marked prospects, good and bad.

Audyense works this way: every prospect has a priority score out of 100 with its breakdown, unknown data is shown as unknown, and once you've marked enough prospects as good or not a fit it adjusts the weights from your marks. The details are in how the Audyense priority score works.