RAG, confidence and health

Mode: reference.

The Investment Health dashboard rolls up a project’s registers into a small set of health indicators. Some are derived deterministically from the underlying data, so editing the indicator directly cannot game the headline health. Others are manual — the self-report is itself the governance act. This page defines each value and its derivation, on both the project dashboard and the Portfolio.

RAG values

The RAG enum (VALID_RAGS) is:

Value	Meaning
Green	On track.
Amber	At risk.
Red	Off track.
Blue	Selectable status value (e.g. complete). Not emitted by the deterministic derivations below.
None	Not applicable / insufficient data to derive.

Investment Health cards

The dashboard shows five status cards: Overall, Confidence, Scope, Cost and Time. Three (Scope, Cost, Time) derive deterministically from the pillars below. Risk has no card of its own — the Confidence card carries the risk evidence (open and materialised counts with a severity mix bar). The Overall card is manual.

Pillar	Shown as	Source	Derivation
Cost	Cost card	Budget totals	Deterministic — see below.
Schedule	Time card	Work packages	Deterministic — see below.
Scope	Scope card	Change register	Deterministic — see below.
Risk	Evidence under the Confidence card	R/I/O register	Deterministic — see below.
Overall	Overall card	PM	Manual — user-selected, not computed from the others.

Rating reasons

The two manual cards (Overall and Confidence) each carry an optional free-text reason, shown on the card beneath the rating.

Field	Card	Notes
overallReason	Overall	Optional. Capped at 120 characters (STATUS_REASON_MAX — enforced in the editor and clamped again on save). '' clears it.
confidenceReason	Confidence	Optional. Same cap and clearing behaviour.

The audit trail records both on change, like every other status field (Overall rating reason / Confidence reason). They annotate a manual call. Nothing derives from them.

Note

During Initiation, derived pillars mute their RAGs until baselines are locked — pre-baseline variance counts as setup, not drift.

Cost RAG

Derived from project budget totals (deriveCostRag).

Result	Condition
Green	forecast ≤ baseline × 1.05
Amber	forecast ≤ baseline × 1.15
Red	forecast > baseline × 1.15
None	no baseline set (baseline ≤ 0 or forecast not finite)

Schedule RAG

Derived from work packages that have both a planned end date and a forecast end date (deriveScheduleRag). Slip = forecast end − planned end, in days.

Result	Condition
Green	no work package slips (max slip ≤ 0 days)
Amber	any work package slips by ≤ 15 days
Red	any work package slips by > 15 days
None	no work package has both dates set

Scope RAG

Derived from the change register over the current reporting period (deriveScopeRag). The period length comes from the project’s reporting cadence.

Result	Condition
Green	zero Approved changes in the current period
Amber	1–2 Approved in the period, or any Proposed change older than 14 days
Red	3 or more Approved in the period
None	no changes recorded

Risk RAG

Counts open downside Risks whose impact × likelihood ≥ 15 (deriveRiskRag — the same threshold behind every “high-impact risk” count). Opportunities (upside) are excluded — they belong on the R/I/O exposure card and lift Outcome confidence, not the Risk RAG. Items with no direction (and legacy Threat records) are treated as downside Risks.

Result	Condition
Green	0 high-impact open Risks
Amber	1–2 high-impact open Risks
Red	3 or more high-impact open Risks
None	no Risks recorded at all

Health on the Portfolio

The Portfolio’s investments table shows one health verdict per investment. It is not a sixth derivation — it reports the PM’s Overall rating from the project’s Investment Health dashboard:

Verdict shown	When
Off track	the PM set Overall to Red
At risk	the PM set Overall to Amber
On track	the PM set Overall to Green

The two surfaces agree by construction. The Portfolio repeats the judgement the PM declared — it does not compute a second opinion.

When the investment is unrated

If Overall is unset (None), the Portfolio falls back to a read of the data — the worse of cost performance and Outcome confidence:

Input	Off track	At risk	On track
Cost variance vs baseline	> 15% over	> 5% over	≤ 5% over
Outcome confidence	Low	Modest / Unknown	High

The row’s action line names which one it is. An investment the PM rated Red carries the PM’s reason. An unrated investment reading Red says so plainly, and asks for a rating.

Note

Two edges to know:

• Blue (Done) is not one of the three verdicts, so a Blue investment falls back to the derived read above.
• No cost baseline counts as 0% variance, so the fallback can read On track while the Cost card shows None.
• The variance cell itself reads No baseline rather than 0%.
• Set baselines and both surfaces agree.

The same verdict drives the off track / at risk counts in the steering headline, and the worst-case rollup on each Theme or Stage group.

Confidence axes

Confidence is self-reported by the PM (manual), additive to RAG. The PM rates two axes, each on the same enum (VALID_CONFIDENCE):

Axis	Question it answers
Outcome confidence	Will this still deliver the outcome we funded it for?
Adoption confidence	Will the change actually be adopted?

Confidence value
High
Modest
Low
Unknown

Note

Confidence is manual by design. The act of a PM declaring confidence is the governance act — Tollgate does not calculate it from the registers.

Reporting-cadence staleness

A project status is stale when its last update falls outside its reporting cadence (isStatusStale). A missing or unparseable last-updated timestamp counts as stale.

Cadence	Period
Daily	24 hours
Weekly	7 days
Monthly	30 days

Staleness drives the dashboard’s stale-update banner and the portfolio overdue-reporting check (see get-portfolio-overdue). Cadence is a recommendation, not a stipulation — the signal surfaces on the dashboard. Tollgate emails nothing.