Deal Health Scoring and Stage Definition-of-Done Automation: A Score That Explains Itself

A dashboard full of red and yellow labels becomes background noise the moment nobody can tell WHY a given deal is flagged — the natural response to an unexplained warning is to stop trusting it, not to act on it. The harder, more useful version of a health score does not just classify a deal; it shows its work, so a manager glancing at a "Critical" tag can immediately see that this particular deal has had no activity in three weeks and is missing its stage's required close date, rather than having to reopen the record to guess.

A believable health score needs real, defensible weights — deciding that a stale deal loses more points than one missing a close date is itself a judgment call that has to be made deliberately and consistently, not derived from a single boolean. It also needs to be stage-aware: what counts as "not ready" at one stage of a pipeline is different from another, which means the underlying requirement list has to be configurable per stage, not a single global checklist. And a score without a reasons list is only half a feature — computing which specific conditions applied to THIS deal, in plain language, alongside the number, is the part that actually makes a health score something a person trusts and uses.

A manager or admin configures, per pipeline stage, which fields matter for a deal to be considered genuinely ready at that stage — a linked company, a primary contact, a non-zero value, an expected close date, or at least one logged activity — and separately decides whether missing each one should be BLOCKING (preventing the deal from advancing to the next stage at all, the mechanic behind Orbit's quality gates) or simply a health-score factor that costs points without stopping movement.

For every open deal, the health score starts at 100 and subtracts a fixed penalty for each condition that actually applies: more than 7 days since the deal was last updated (-30, the single largest factor), no logged activity in the last 14 days (-20), the current stage's configured requirements not fully satisfied (-20, regardless of whether any individual requirement is marked blocking), no expected close date set (-15), a deal value of exactly zero (-15), and low priority combined with already exceeding the stage's own stale-days threshold (-10). The result is floored at 0 (never negative) and mapped to one of three labels: 70 and above is "Strong," 40 to 69 is "Attention," below 40 is "Critical." Alongside the number and the label, a parallel array of plain-language reasons lists exactly which of the six conditions actually triggered for that specific deal — a deal at 55 because of staleness and a missing close date shows both reasons explicitly, never just the number.

A sales manager reviews the pipeline and sees two deals both labeled "Attention." The first shows a score of 55 with reasons listing "no update in 7+ days" and "no expected close date" — the deal has had recent conversation but nobody ever set a target date. The second shows a score of 65 with a single reason, "current stage requirements not met" — this pipeline's "Proposal Sent" stage requires a logged activity and an assigned value, and this deal has neither, even though it was updated yesterday.

Without the reasons list, both would just read "Attention, deal with it" — with it, the manager immediately knows the first deal needs a close-date conversation and the second needs the proposal's value actually entered, two completely different next actions from the same label.

A manager scanning a pipeline gets a consistent, weighted signal instead of a subjective sense of which deals "feel" behind, and — critically — gets the specific reason without opening every flagged deal individually. Because the underlying requirements are configured per stage, the score reflects what actually matters at each point in a specific pipeline, rather than one generic checklist applied everywhere.

A team that has not configured any per-stage requirements will still get a score, but the "stage requirements not met" factor will never apply for stages with nothing configured — the score is only as stage-aware as the configuration behind it. A team looking for the score to predict deal outcome probability (a true forecasting model) will not find that here — this is a transparent, rules-based health indicator built from named factors, not a machine-learning win-probability estimate.

A simple status flag tells a manager something is wrong without saying what, which either trains people to dig through every flagged deal manually or, more commonly, to stop trusting the flag at all. Centriu Orbit's health score keeps the at-a-glance simplicity of a tier label while attaching the specific, named reasons behind it — so acting on a flagged deal starts with knowing exactly what to fix, not with reopening the record to guess.