Pick the Right Decision Shape

GTM Lab's sales leadership asks for “a health number on every open deal.”

What gets built is a score from 0 to 100 that blends how recently the buyer was in a meeting, how long the deal has sat in its current stage, how many of the qualification criteria are filled in, and how big the deal is. One number, on every record, sortable.

Within a month it is being used to answer four different questions. Is this deal moving? Is this deal qualified? Is this deal worth my time this week? Should this deal be in the forecast? It cannot answer all four, because they are not the same question, and a single number that tries produces a value where a high score can mean “large deal with nothing happening” and a low score can mean “small deal closing on Friday.” Nobody notices, because both numbers look fine.

Four shapes, four different jobs

Classification puts a record into one of a closed set of categories. Use it when the answer is a kind, and the consequence differs by kind. Inbound qualification is a classification.

Scoring puts a record on a scale. Use it when the answer is a degree and you need to compare records against a threshold or against each other. Scores are seductive and are the shape most often chosen wrongly, because a number feels more precise than a category and usually is not.

Ranking orders a set. Note the difference from scoring: a rank is only meaningful relative to the rest of the set, and it changes when the set changes without anything about the record changing. A rep asking “what should I work today?” wants a ranking. A manager asking “is this deal healthy?” does not, and giving them a rank will mislead them every time the population shifts.

Recommendation proposes an action. It is the only shape that carries an implicit “and you should do something about it,” which is why it needs the tightest action limits.

One question per shape

The fix for the opening is not a better formula. It is two numbers. GTM Lab ends up with an engagement score answering “is this deal moving?”, built entirely from activity dates, and a separate qualification grade answering “is this deal qualified?”, built from the qualification criteria. A deal can score high on the first and low on the second. That combination is not a contradiction to be smoothed away; it is the most useful thing the two numbers can tell you, and a blended score destroys it.

The rule is worth stating plainly: if a number would change for two unrelated reasons, it is two numbers.

Decay, not windows

One design detail belongs here because it decides the shape's behavior more than the formula does. The obvious way to turn a date into a score is a window: a meeting in the last 30 days counts, older than that does not. This produces a cliff. A deal is healthy on day 29 and unhealthy on day 31, nothing happened in between, and every report that samples on a Monday tells a different story from one that samples on a Friday.

Decay instead. A meeting is worth the most on the day it happens and less every day after. There is no cliff, the number moves the way the underlying reality moves, and a stale deal drifts downward visibly instead of falling off a shelf.

Carry this into your business

Take a score your team already uses and ask what two different things could each move it. If you can name two unrelated ones, you have found a number that is quietly answering a question nobody asked, and splitting it is usually a week of work that saves a quarter of arguing.