How to
Back testing a deal score against your own closed deals.
To back test a deal score, pick a fixed point before each closed deal ended, work out what the score would have been from only the activity that existed by then, and compare the average for won deals against lost deals. Then look at which signals drove the gap.
The four steps.
Pull every deal you have closed, with its outcome, its close date and the activity attached to it: emails, meetings, notes, stage changes and contacts added.
Choose one point shortly before the close and use it for every deal, then use everything that happened up to that point. A long deal is defined by its whole history, not by its last few weeks. The point itself keeps the outcome out of the score.
Score each deal using only what had happened by that point. Ignore everything after it.
Average the scores for wins and for losses, then compare them.
Why a fixed point matters.
If each deal is scored at a different moment relative to its close, the comparison mixes deals that were nearly done with deals that were early. A fixed point makes them comparable and stops late activity from flattering the wins.
The step after the comparison.
The gap tells you whether the score works. It does not tell you why. For that, go signal by signal and ask how often each one appeared on wins and how often on losses. The ones that appear far more on wins are your real predictors. The ones that appear evenly on both are noise, however sensible they sound.
Reweight the score around the first group and repeat the test. The gap should widen. If it does not, the weights are still wrong.
Doing it without building it.
For a few hundred deals this is a spreadsheet job. Beyond that it is a data job. Kaypo runs it automatically against a connected CRM or an export of closed deals, reads every closed deal you have, and reports the gap, the signals that mattered and the ones that did not. It is free at kaypo.io/hindsight.
Questions
What is a back test in sales?
A back test checks a prediction against outcomes that are already known. In sales it means scoring deals you have closed as they stood before they ended, then seeing whether the scores separated the ones you won from the ones you lost.
Why score a deal before it closed rather than after?
Activity after a deal closes, such as onboarding and kickoff meetings, is caused by the outcome, so counting it flatters every win. Kaypo scores each deal on everything up to the point it closed and nothing after, and uses the same point for every deal so the comparison is fair.
How far back should a back test look?
Use every closed deal you have. Older deals are not less informative about how your buyers behave. If you changed your product, pricing or target customer, compare recent deals with older ones and see whether the signals moved.
Can I back test a deal score in a spreadsheet?
A spreadsheet works for a few hundred deals if you can export dated activity with each one. Beyond that, or when the activity lives in emails and meetings, an automated run is far less error prone.
It arrives in Slack
Every morning, each rep gets one message naming the deals on their own book that moved overnight, biggest mover first, with the reason beside each number and the deals that have gone quiet underneath.
$99 a month, cancel any time.. Everyone on the team. No per seat charge. Connect a CRM and it runs. No implementation fee. 14 day trial. No card.
Related reading
Checking whether your deal score is any good
A deal score is only accurate if the deals you won scored higher than the deals you lost, before they ended. How to check yours in an afternoon.
Running a win loss analysis from what is already in your CRM
You do not need interviews to learn why deals close or die. Most of the answer is in the activity your CRM already holds. How to pull it.
A free back test of your deal score, on your own closed deals
Kaypo Hindsight is a free, read only back test. It reads your closed deals and reports what separated the ones you won from the ones you lost.
What actually predicts whether a deal closes
Every scoring tool asserts which signals matter. The only way to know is to check them against the deals you already closed.
Reviewed September 20, 2026 against the product as it behaves today.