Sales prompt of the day

Build the fit scorecard from your own closed-won and closed-lost

Everyone has an ICP slide and nobody built it from data. Derive the account-fit scorecard from the deals you actually won and lost.

You are a RevOps analyst deriving an ideal-customer profile from closed deal history. I will paste won and lost accounts with their attributes.

Produce:

A) THE DISCRIMINATORS — attributes in MY data that separate won from lost, each with the win rate on each side and the sample size behind it. Any attribute resting on fewer than ten deals per side is labeled "too thin to score" and excluded from the scorecard rather than quietly weighted.

B) THE SCORECARD: the surviving attributes as a simple point system a rep can apply in under a minute, with the score bands and what each band means for routing — prioritize, work opportunistically, or disqualify.

C) THE BACKTEST AND THE BLIND SPOTS: how this scorecard would have rated my pasted deals (including the wins it would have wrongly disqualified), plus the attributes I did not give you that most likely matter and how to start capturing them.

Inputs: [CLOSED-WON ACCOUNTS: INDUSTRY, SIZE, TECH, SOURCE, DEAL SIZE, CYCLE LENGTH] · [CLOSED-LOST ACCOUNTS: SAME FIELDS + LOSS REASON] · [ANY CURRENT ICP DEFINITION]

Rules: Do not invent attributes, win-rates, or deals I did not give you, and never report a discriminator as meaningful on a thin sample — label it "too thin to score". Keep confidential CRM, deal, and customer data out of consumer AI tools and follow your employer's AI-use policy. This proposes a scorecard; you validate it against a holdout period before it routes real leads. Verify the surviving discriminators against your CRM's own reporting before anyone routes a lead on this scorecard.

Why it works — ICP definitions are usually written from the three customers leadership likes to name, which is how a team ends up disqualifying its own best segment. Deriving discriminators from won AND lost accounts is the only honest version, and forcing the backtest to list the wins the scorecard would have thrown away is what keeps a tidy model from quietly shrinking the market.