Deal Risk Scorer & Early-Warning
Score every open deal's risk from the signals you choose, explain each score, and surface the shaky deals early so your team saves the ones that matter.
Import deals and a versioned risk rubric, compute an explained risk score for each deal, bucket high/medium/low, let a manager review and override before publishing, then export a scored CSV and email an at-risk digest.
Before you start
- A deals export (CSV or Google Sheet) with deal id, stage, amount, close date, last activity date, and contact count
- A risk rubric you can describe (signals, weights, thresholds)
- [object Object]
The problem this kills
Your forecast says these deals are committed. Your gut says half of them are quietly dying. The deal sits in the same stage for three weeks, the rep talks to one contact who stopped replying, the close date has slipped twice, and the discount keeps creeping up - but nobody connects the dots until the quarter is already blown.
Most teams either eyeball the pipeline deal-by-deal (slow, subjective, and different every manager) or trust a CRM "probability" number nobody can explain. Reps don't trust a black-box score, so they ignore it. Managers can't coach against it, because it doesn't say why a deal is risky.
You need an objective, repeatable read on which committed deals are actually shaky - one that shows its work, so reps believe it and managers can act on it.
What you'll build
A small internal web app where you:
- Import your open deals (CSV or Google Sheet) with the activity, contact-count, and close-date fields you already track.
- Upload a risk rubric you define - the signals that matter to your business (days in stage, days since last activity, single-threaded contact, no next step booked, discount depth, close-date pushes), each with a weight and threshold.
- Get an explained risk score per deal - not just a number, but the exact signals that pushed it up, so reps trust it.
- Bucket every deal into high / medium / low risk tiers.
- Review and override any deal's tier as a manager, with a logged reason, before anything is shared.
- Publish the risk list, export a scored CSV, and send an at-risk digest email to the right people.
- Every high-risk deal comes paired with a suggested rescue play so the team knows what to do next.
What's inside the Implementation Plan
- It starts by interviewing you about your business. Before it builds anything, the plan has the AI agent ask you about your real pipeline: your stages, the exact field names in your deals export, how you measure activity, what "single-threaded" means to you, your typical and peak deal counts, and your messy edge cases. It reflects back a short tailored spec, you give a thumbs-up, and then it builds a tool shaped to your data - not a generic template.
- A step-by-step build, each step ending with a ready-to-paste prompt for your AI coding agent.
- A versioned rubric so every score is reproducible - you can always answer "why was this deal high-risk last month?"
- Per-signal score explanations on every deal.
- A manager approval + override gate before the list goes out.
- A "No API yet?" fallback so you can build the whole thing today from a Google Sheet and ship a clean scored CSV - no CRM integration required.
The governance it includes (this is the point)
- Login so only your team can open the tool.
- Row-level security so each organization only ever sees its own deals and rubric.
- A full audit trail - who imported, who scored, who overrode which deal and why, and when.
- A human-in-the-loop gate: the AI drafts the scores, the manager approves the rubric version and can override any deal's tier with a logged reason, and only then is the risk list published to reps.
- Duplicate guards keyed on deal id (with rubric version stamped on every score) so the same deal can't be double-counted or silently re-scored under a different rubric.
Who it's for
Sales managers and revenue operations (RevOps) leaders who want an objective, repeatable read on pipeline health - and reps who'll only act on a risk score if they can see exactly why it landed where it did.
You've got this - paste the first prompt and let the agent interview you.