Question

How many closed deals you need before deal scoring works

About 30 closed deals, with at least 8 wins and 8 losses. What each kind of scoring asks for, what to use before then, and how to tell when you have enough.

The short answer

About 30 closed deals, with at least 8 wins and 8 losses, is enough to start fitting a deal score to your own history. Below that a fitted model mostly memorizes noise, so start from industry benchmarks. Some platforms ask for far more: Gong publishes a minimum of 50 wins and 150 losses from the last two years.

How Kaypo answers this on your pipeline

30 minutes with the founder

See how Kaypo helps your pipeline

Thirty minutes on your pipeline or the example one. Pick a time and it is booked.

What each approach asks for before it scores

KaypoGong deal likelihood scoreIndustry benchmarks alone
Closed deals it needsAbout 30, at least 8 won and 8 lost50 won and 150 lost, from the last two yearsNone
Before you have themScores on industry benchmark weights from day onePublished as the minimum for the scoreThe same weights for every team
How it decides your own model is readyTests it on closed deals it never saw, and keeps the benchmark until yours winsNot stated in the minimumsIt never learns your deals
  1. 1

    Why about 30, and why 8 of each?

    A score learns what your wins had that your losses did not. With a handful of closed deals, one unusual win can look like a pattern, and the model memorizes it. Around 30 closed deals is where a signal has to show up more than once on each side before it counts.

    Both outcomes matter. Thirty wins and two losses teach a model nothing about losing, so Kaypo waits for at least 8 of each. Each deal also needs some activity on it, at least three signals, because a deal with one meeting logged looks the same whether it was won in a week or lost over nine months.

  2. 2

    What should you use before you have 30?

    Industry benchmark weights, applied to what each buyer actually did: replies, meetings held, more people joining, a security or legal review. That is a sensible score on day one, and it is the starting point your own history improves on.

    Start tracking buyer signals from the first deal, so by the time you have 30 closed, each one carries the evidence a model needs.

  3. 3

    How do you know your own model is ready?

    Test it on closed deals it was never shown. If it ranks those wins above those losses more often than the benchmark does, it has learned something real. If it does not, keep the benchmark.

    Kaypo runs that test every night once there are about 30 closed deals, switches only when your own model wins on the held back deals, and says on the Model page which one is in use and why.

  4. 4

    Why do some platforms ask for so many more?

    A model that weighs hundreds of signals per deal, many of them from recorded calls, needs far more history before those weights settle. Gong publishes a minimum of 50 closed won and 150 closed lost deals from the last two years for its deal likelihood score.

    A team that closes a few deals a month can take a year or more to get there. Starting from benchmarks and testing your own fit as it grows gets that team a working score now.

The real product, playing.

Recorded in the sandbox, the same example pipeline you can open yourself. Pick any part.

Deals. Every open deal, scored on what the buyer did.

Questions people ask.

Why does deal scoring need closed losses as well as wins?

A deal score learns what your wins had that your losses did not, so it needs both. Thirty wins and two losses say nothing about how deals are lost, which is why Kaypo waits for at least 8 of each before fitting a score to your own deals.

Can a team with few closed deals use deal scoring?

A team with few closed deals can use deal scoring when the score starts from industry benchmarks and reads what buyers do. Kaypo scores from day one that way, then switches to a model fitted on your own deals once it beats the benchmark on deals it never saw.

How many closed deals does Gong need for deal likelihood scores?

Gong publishes a minimum of 50 closed won and 150 closed lost deals from the last two years, each open at least 3 days, plus the previous quarter of data.

Sources

Read next.

Reviewed November 8, 2026 against the product as it behaves today.

30 minutes with the founder

See how Kaypo helps your pipeline

Bring this question and your own deals. Pick a time and it is booked.

Next step

See it in Deals, on the example pipeline

The real product on invented B2B software companies. Nothing to connect and no signup.

Open Deals in the sandbox

Or back test your own closed deals, free, or talk to a person for 30 minutes.

See Kaypo before you connect anything.

The whole product on an example pipeline: scores, next steps, plays and agents. No signup and no call.

$1,500 a month, or $15,000 a year, for one admin and up to 15 users. Free for 14 days, no card.

Or get the free report on your own closed dealsOr talk it through with a person, 30 minutes