An annual plan as structured, measurable and tracked — pillar to objective to key result to quarterly milestone to recorded actual, with the status derived rather than typed, and direction applied per measure so half the plan is not scored backwards.
Most annual plans are documents. They are approved in March, laid out in prose and a table of ambitions, and then live in a folder while the organisation reports performance against a different set of numbers in a different place. By the time anybody compares the two, the year is over.
commvita treats the plan as structured, measurable and tracked: a plan holds sections, each section holds outcomes, and every outcome is a key result with a unit, a direction, a baseline, an annual target and quarterly milestones. The actual is recorded against the milestone, and the status is derived from the two.
Half the measures a health system cares about get better by going down — waits, delays, readmissions, response times — and half by going up. A single up-is-good rule scores half a plan backwards, and it does so quietly, because a red that should be green still looks like a plausible red.
Derived, not typed. The outcome's current status is computed from the recorded actual against its milestone target. Nobody sets it by hand, so the status on a board paper cannot drift away from the number printed next to it — which is the commonest way a performance report becomes internally inconsistent without anyone editing anything.
An outcome carries three fields that turn a list of targets into a strategy: the objective it is a key result for, the strategic pillar it serves, and a delivery confidence from nought to a hundred. All three are nullable, so a plan written before they existed is unaffected and nothing had to be back-filled with a guess.
The pillars are deliberately generic — P1 to P5, and the shipped set (Better Outcomes, Better Access, Prevention First, Thriving Workforce, Financial Sustainability) is written in universal healthcare language. Baking one nation's planning framework into the data model would make the module unusable everywhere else and would go stale in the nation it came from the first time the framework was reissued. An organisation names its own pillars; commvita holds the structure.
An outcome that ends the year at 74% against an 85% target has failed. The useful question is when that became knowable — and a plan that only records an annual figure cannot answer it. Milestones are held per quarter, actuals are recorded per period, and the trajectory is the object a board actually needs.
| Part | Where | Status |
|---|---|---|
| Plans, sections, outcomes, actuals | /annual-plans |
Live |
| API | /annual-plans — 12 endpoints |
Live |
| Outcome tracker, OKR trajectory, strategic alignment | GET /annual-plans/{id}/outcomes |
Live |
| Multi-year trajectory by outcome code | GET /annual-plans/trajectory/{org} |
Live |
| Plan comparison | GET /annual-plans/compare |
Live |
| Jurisdiction-agnostic reference plan (P1–P5, 14 outcomes) | seeded | Demonstrated |
Actuals are recorded, not harvested. An outcome carries the data source it should come from and, where one exists, a link to a platform module — but the value itself is entered. commvita does not currently compute a plan actual from the underlying operational record, so an outcome is only as current as the last time somebody recorded a period against it. An outcome with no actual reads as grey, which means not yet recorded and must not be read as on track.
A stored actual may carry a supplied status. The outcome's own status is always the derived one. The individual actual row will accept a status supplied by the caller, so a historical row loaded from another system keeps the rating it was given there — useful for migration, and worth knowing when reading a single row rather than the outcome.
What this is not. A planning and tracking record, not a forecasting engine. It does not predict whether a target will be met, does not model the effect of an intervention, and does not decide priorities. Delivery confidence is a judgement a named person records, not a computed probability.