Anatomy of a Deal AI Health Score: What Every Sales Leader Should Look Inside

Imagine a £140k deal in 'Proposal' that is scoring 70. The headline number is reassuring. The breakdown is what changes the conversation. Engagement velocity contributes 22 of the 70 (strong — buyer is replying inside 24 hours, meeting cadence is weekly). Multi-threadingHealth · GlossaryMulti-ThreadingThe practice of building active relationships with three or more buying-committee contacts inside a target account. Single-threaded deals (one champion only) are 3–4× more likely to slip when that contact leaves, goes silent, or loses internal political capital.View full definition → depth contributes only 8 of a possible 20 (weak — one championHealth · GlossaryChampionThe internal contact at a target account who actively advocates for your solution to the wider buying committee. Losing a champion mid-cycle is one of the strongest leading indicators of a deal slipping or dying outright.View full definition →, no exec sponsor). Stage-fit contributes 18 of 20 (typical proposal duration). Deal shape contributes 22 of 30 (close fit to recent £100–200k closed-won deals).

That breakdown tells you exactly what to do this week: book an exec sponsor introduction. The headline 70 would have led to complacency. The component view leads to a specific play that lifts the score from 70 to 85 inside ten days — the difference between a likely close and an inevitable one.

Healthy (75–100): close-rate ≥ 70% in your historical cohort. Lock these in, don't over-invest. Monitor (50–74): close-rate 35–55%. This is where most coaching ROI lives — small interventions move the needle. At-Risk (25–49): close-rate 10–25%. Triage decision required: rescue with an exec sponsor, or remove from forecast and reclaim the rep's time.

Critical (0–24): close-rate under 10%. Stop forecasting these deals immediately. Counterintuitively, the cleanest signal of a healthy organisation is a forecast that excludes Critical deals entirely — even if the rep is still working them. Hope is not a forecasting input.

The biggest mistake leaders make is to treat the score as the answer. The score is the second opinion. The rep has context the model does not — the procurement contact who hates Mondays, the CFO who is about to be replaced, the legal team that just rejected a similar contract. Both perspectives matter.

The cleanest operating model is to inspect every deal where the rep call and the model disagree by more than 20 points. If the rep can articulate the missing context, trust the rep and adjust the forecast manually. If the rep is just optimistic, trust the model. Within two cycles, this discipline closes the gap because reps internalise the signals the model is reading.

A score that just says '34/100' is unfalsifiable. Reps cannot push back. Coaches cannot coach. Boards cannot interrogate it. A score that breaks into engagement, structure, fit and time-in-stage gives every stakeholder a way to engage with the model — to argue with it, to learn from it, to override it when they have better information.

Explainability is not a 'nice to have' for AI in sales. It is the precondition for adoption. Tools that hide their reasoning get ignored within a quarter. Tools that show their working get used every day, by every rep, on every deal — which is the only way the value compounds.