A lead scoring model that earns its keep

A lead score is not a grade you award a lead. It is a bet you place on your own time, and most B2B scoring models quietly lose that bet because they collapse two genuinely different questions into one tidy number. A composite of 72 tells you almost nothing useful: it could be a perfect-fit VP who has not yet moved, or it could be a student who happens to have read every blog post you have ever published. Both arrive at the same number, and both demand opposite responses, so the moment you act on the average you are already wrong.

The fix is to stop averaging and start separating. Score fit and intent as two distinct axes and let each drive its own decision. Fit, the firmographic and demographic match, decides whether a lead deserves your attention at all, so it works as a gate. Intent, the behavioural and buying signal, decides whether now is the moment, so it works as a timer. Route on the pair, never on a blended score, and the model starts telling you something a single figure never could: who to chase, who to nurture, and who to ignore even though they look busy.

This matters far more when you are the founder, because you are the SDR, the AE and the marketer at once, and a bad bet does not cost you a misallocated rep, it costs you the only hour you had today. A model earns its keep only when it measurably routes that scarce attention toward the leads most likely to convert, and you prove that with a conversion-lift number against an unscored baseline, never with the elegance of the point table. By the end of this guide you will be able to build a two-axis model in an afternoon, derive your own thresholds from your own data, wire it to a speed-to-lead trigger, and decide, with evidence, whether it deserves to keep running.