Question
What actually predicts whether a deal closes.
Most deal scoring applies an industry template: replies are good, meetings are good, silence is bad. Those assumptions are right for some businesses and wrong for others, and the only way to tell which is to run them against your own closed deals and see which ones actually separated your wins from your losses.
The assumptions almost every tool ships with.
A reply means interest. A meeting means progress. A security review means the deal is real. Silence means it is dying. Those are reasonable starting points and they are guesses.
They are also the same guesses for every customer of that tool, which cannot be right, because a security review is routine in one business and a six week stall in another.
What checking them looks like.
Take every deal you closed, won and lost. Score each one using only what was known before it closed. Then ask whether the winners scored higher than the losers.
If they did, the model works on your pipeline. If they did not, the model is describing somebody else.
The two week gap matters. The run up to a close is full of activity caused by the close, and a model that reads it is marking its own homework.
The findings that surprise people.
More meetings often predicts losing. Churning through demos is a symptom of nobody being able to decide rather than a sign of interest.
Reply speed frequently predicts nothing at all. Polite fast repliers lose deals at the same rate as slow ones.
A security review predicts almost nothing on its own, and predicts winning strongly when somebody has also been named as the decision maker. Alone it is a process. With an owner it is a process going somewhere.
Why this cannot be a template.
A model fitted to your closed deals is fitted to your sales motion, your buyers and your pricing. The same fit on another company produces different weights, and that is the point rather than a flaw.
The honest version of a deal score is one that can show you which of your own outcomes it learned from.
How Kaypo produces this.
Kaypo scores every open deal, then runs the same weights back across the deals you have already closed and reports whether your wins scored higher than your losses. When your own history disagrees with the benchmark, the Fit page names the signal and the gap.
Questions
How many closed deals does this need?
Enough that a handful of them changing does not change the answer. Under about forty, most signals correctly come back with no verdict and a benchmark is used instead.
Does it need intent data?
Deal scoring does not need intent data. The strongest predictors are usually things already in the CRM: who was named, what stage it reached, whether anybody replied, what was said on a call.
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
How to decide which deals to work today
With forty open deals and a day, the question is not which deals are biggest. It is which ones your attention actually changes.
How to tell which of your open deals are actually real
Every pipeline has deals that are alive and deals being carried. The difference shows in what the buyer did, not the stage somebody dragged the card to.
How to score deals in Salesforce or HubSpot
The three ways to score deals in a CRM, what each costs in time and money, and why most home built scoring models stop being maintained within a quarter.
The questions worth asking in a deal review
Twelve questions that surface whether a deal is real, organized by what they are actually testing.
Do more meetings mean a deal is more likely to close?
Meeting count is one of the most trusted sales signals and one of the least tested. Why it can point the wrong way, and how to check yours.
Reviewed September 20, 2026 against the product as it behaves today.