Designing a scoring rubric that survives Q4 calibration
How to write a lead scoring rubric simple enough that your AE will actually use it, and durable enough to outlast a quarter end.
Every lead scoring system I have ever seen falls apart between Halloween and Christmas. It is a small annual tragedy. The model was tuned in February when pipeline was thin and any signal counted. By November the team has learned the model is wrong in a hundred small ways. Reps quietly stop trusting the number. By January, leadership wonders why the dashboard says 80 percent of leads are A grade and nobody is closing.
The bug is not the model. The bug is the rubric.
Most rubrics fail because they try to capture too many dimensions at once. Industry fit, persona fit, intent strength, geographic fit, technographic fit, timing fit, all in one number from 0 to 100. The result is a score that nobody can defend in a deal review. Defensibility is what makes a rubric survive a calibration cycle. If the AE cannot explain why a lead is a 78 in plain English, the AE will not trust it in November.
The four-question rubric
We have moved every rubric we touch toward the same shape. Four questions, each answered yes, no, or maybe. The composite score is just a function of those four. Anything more complicated and the AE stops reading it.
- Does this account actually buy the category of work I sell? (Yes, no, or maybe.)
- Is the trigger time-sensitive in a window of 30 to 90 days? (Yes, no, or maybe.)
- Is the project value within the band I can do profitably? (Yes, no, or maybe.)
- Do I have a path to the actual buying human, or just a directory? (Yes, no, or maybe.)
Why this survives Q4
Because the four questions are stable. The market changes. The buyer changes. The trigger windows shrink and grow with the quarter. But the four questions stay legible. When a rep complains that a 92 turned out to be a dud, you can ask which of the four questions was wrong, and the answer points you straight at the calibration. You are not arguing about a black box. You are arguing about one of four sentences.
If the score on your dashboard does not decompose into a sentence the rep can finish, you do not have a rubric. You have a number.
What we recalibrate, and when
We recalibrate the thresholds three times a year. End of February, end of June, end of October. We never recalibrate in December because nobody has clean enough data after Thanksgiving. We change the weights on the four questions, but not the questions themselves. The questions are the contract with the team. The weights are the science.
“A scoring system that requires a data team to defend is a scoring system the sales team will stop using.”
If this resonated, it'll feel familiar in the product.
Try Blacksmith against your real territory for 14 days. No card, no metered AI credits, no surprises.
Summit seats are limited and allocated to qualifying founding members. Perks subject to final terms.