Deal Health Score: Definition, Components & How It Predicts Close Probability

Key Takeaway: Deal health score is a composite 0-100 score that predicts the probability of an open opportunity closing, based on engagement signals (recency, multi-thread coverage, executive presence) and qualification completeness (MEDDIC or BANT dimensions populated). It is to open pipeline what predictive lead scoring is to top-of-funnel leads.

What is a Deal Health Score?

A deal health score is a machine-learning-derived probability indicator assigned to each open opportunity in a CRM. It answers one question continuously: given everything the system knows about this deal right now — who has engaged, when, at what level, how completely it is qualified, and how it compares to historical deals that closed or churned — how likely is it to close?

The score is expressed as a 0-100 number (or equivalent probability), updated in near real-time as new signals arrive. It is not the AE's gut feel encoded as a field; it is a model output derived from primary-source data.

The business case is simple: pipeline reviews based on AE self-reporting are systematically optimistic. Reps report what they believe (or what they hope). Deal health scores report what the data shows. The gap between the two — the "happy ears" delta — is the risk that forecasting models built on self-reported data systematically underestimate.

Core Signal Categories

Engagement recency and velocity. When did the last meaningful two-way interaction occur? A deal where the prospect opened an email 3 weeks ago and the last call was 5 weeks ago scores lower than one where there was a call 4 days ago followed by a proposal request. Recency is the single highest-weight signal in most models because it correlates strongly with deal momentum.

Multi-thread coverage. How many distinct stakeholders at the account have been engaged? A deal with only one contact engaged is single-threaded and scores materially lower than one with three or more committee members active. Single-threaded deals are the highest-risk category in enterprise pipelines. See Buying Committee Detection for how committee members are identified.

Executive presence. Has a senior stakeholder (VP or above) been active in the deal — attending a call, replying to an email, requesting a business case? Executive engagement is a strong close signal because it indicates organizational priority.

Qualification completeness. How many MEDDIC (or BANT) dimensions have confirmed evidence? A deal in Stage 3 with no identified Economic Buyer, no documented Decision Process, and no quantified Metrics is a structurally weak deal regardless of AE optimism. See MEDDIC AI for how qualification completeness is extracted automatically.

Stage velocity. How long has the deal been in its current stage relative to the historical median for deals of similar size and segment? A deal in Stage 2 for 60 days when the historical median is 14 days is flagging a stall.

Competitive signals. Has a competitor been mentioned? How did the rep respond? Deals with active competitive threats and no documented competitive response are higher risk.

How to Use It for Forecasting

Deal health scores aggregate from individual opportunity to portfolio in two ways.

Weighted pipeline. Replace the binary "included in forecast" / "excluded from forecast" with score-weighted expected value: deal value × close probability (derived from health score) = expected contribution. A €500k deal at health score 30 contributes €150k of weighted pipeline; a €200k deal at health score 85 contributes €170k. The weighted total is a more accurate forecast than the sum of deals the AE included with an optimistic confidence level.

Cohort risk alerts. Segment the pipeline by health score cohort (0-40 = at risk, 41-70 = requires attention, 71-100 = healthy). Set automated alerts: any deal above a defined deal size that drops into the at-risk cohort triggers a manager review flashcard. The agentic layer can propose specific recovery actions — schedule an executive call, send a competitive response, request a procurement intro — based on which signals are weakest.

Vendor Implementations

Gong Forecast. Derives health signals from call analysis and email activity; strong on engagement recency and competitive signals.

Clari. Aggregates CRM activity, email, and calendar data; strong on multi-thread and stage velocity; known for its forecast accuracy at portfolio level.

Salesforce Einstein Opportunity Scoring. CRM-native; relies on CRM activity data; weaker on signals that live outside Salesforce (call recordings, LinkedIn).

Knowlee 4Sales. Stores health signals as graph properties on deal nodes, making health score components queryable and auditable. The agent layer uses health score drops as triggers for automated recovery recommendations surfaced through the human-in-the-loop approval flow on the operator dashboard.

Related Concepts