Someone asks for the RAG. You look at a late milestone, a thin contingency and a sponsor who wants “green for the pack”. Ten minutes later the slide is green. The plan is not.
RAG is a colour summary of project health or delivery confidence against agreed expectations. Teams use thresholds, evidence-informed professional judgement or both. This guide uses InsightSpotter’s chosen scorecard method: explicit evidence rules, a conservative worst-of roll-up for agreed material dimensions, logged overrides and a road to green for non-green results. It gives you starting thresholds, a one-page Excel-style calculator specification and a short SteerCo narrative.
Definition — RAG status criteria: the agreed meanings and assessment criteria for Red, Amber and Green. They help explain project health and the attention or decisions required. Computable rules and worst-of aggregation are choices in this guide, rather than requirements that define all good RAG reporting.
As of October 2026, this is an editorial scorecard method for local adaptation, not an industry scoring standard or a validated prediction model. Agree its dimensions, thresholds, evidence rules, authority and review cadence with your board or PMO. A repeatable calculation can still use the wrong baseline, an optimistic forecast or an unsuitable rule.
If reporting itself is the bottleneck, pair this with how to reduce project reporting time. If money is the driver, see what happens when a project goes over budget.
Most PMOs already invent their own RAG meanings and thresholds. There is no single industry scoring standard. InsightSpotter publishes this scorecard for education and local adaptation: clear bands, evidence rules and a worst-of roll-up you can copy, challenge or replace. Treat the numbers as teaching examples, not a methodology you must adopt.
For a neutral definition of project governance (authority, escalation and reporting), see the APM overview of governance. UK public-sector teams can also check expectations in the Government Functional Standard GovS 002: Project Delivery.
What RAG Is For (And What It Is Not)
RAG (Red / Amber / Green) is a status signal for decision makers. It is not a forecast, not a risk score on its own and not a substitute for the underlying schedule, cost and issue records.
- Green: within agreed tolerances; no escalation required this cycle under the selected rules.
- Amber: outside at least one soft threshold, or on a path that will breach soon without action; management attention needed.
- Red: outside hard tolerance, or recovery needs a decision the project manager cannot take alone (scope, money, date, risk appetite).
- Not assessed: a required assessment cannot be supported. This is a separate state, not a fourth colour or a quiet Green. Open adverse exceptions remain visible under the rules below.
In PRINCE2 language, RAG should line up with management by exception: you escalate when tolerances are forecast to break, not when a slide needs a warmer colour. Your SteerCo pack should say what you need—not only how you feel. See also steering committee basics.
Starting Thresholds You Can Type Into Excel
The bands below are example starting points for a single project reporting weekly to a SteerCo. Agree them in writing. Change them for portfolios, regulatory programmes or fixed-date events. Do not treat them as industry averages—they are not.
Choose the material dimensions before the cycle and mark every included dimension as required. Record any agreed exclusion, its rationale and approver; an excluded dimension is not Green. Do not drop a row to improve the result or hide an open exception. Where mandatory acceptance matters, include quality as a required dimension.
Apply precedence before formulas: retain unresolved adverse exceptions first; check whether current inputs and rules are usable; then test Red before Amber before Green. Where a retained exception and a current assessment coexist, use the worse colour for that dimension and show any evidence gap separately. If evidence is unusable and there is no retained exception, use Not assessed. Missing information is never the default Green branch.
| Dimension | Input and basis | Green | Amber | Red |
|---|---|---|---|---|
| Time | Signed calendar days late, d = forecast finish minus approved baseline finish, for the same named project finish or material milestone | d <= 5, including on-time and early forecasts | 5 < d <= 15 | d > 15 |
| Cost | v = (EAC – approved budget) / approved budget; same currency and agreed contingency basis | v < 5%, provided the Red contingency condition does not apply | 5% <= v <= 10%, provided the Red contingency condition does not apply | v > 10%, or agreed contingency is exhausted and further overspend against the approved budget is forecast |
| Scope / change | Approved scope record, current build and material open changes | Build matches the approved baseline; no material change awaits decision | Material change awaits decision, but delivery still matches the approved baseline | Unapproved scope is in delivery, or the baseline no longer matches what is being built |
| Risk / issue exposure | Current residual delivery exposure and open issue severity, using the project’s agreed scale and appetite | No high or critical item; residual exposure is within agreed appetite and no issue threatens the next material milestone | At least one high item, exposure outside agreed appetite, or an issue threatening the next material milestone, without a Red condition | Any critical item, or one or more high items blocking the next material milestone |
| Benefits (if tracked) | Evidence that assumptions and the benefits case remain credible | Key assumptions hold, or an agreed replacement supports a credible benefits case | A key assumption is invalidated and replacement is unresolved, or a reset path is being assessed; the Red condition does not apply | Benefits case is no longer credible and there is no reset path |
| Quality (where required) | Required acceptance checks and their documented results | Required checks due by this cycle passed; no open mandatory failure | An acceptance concern requires action but no mandatory check has failed | A mandatory acceptance check failed and has no accepted resolution |
Exact boundaries and usable inputs
Time uses calendar dates at the same reporting cut-off. Strip time-of-day values and subtract the baseline date from the forecast date: the baseline day is zero; the next calendar day is one day late. Weekends and holidays count. Exactly 5 days late is Green; exactly 15 is Amber; 16 is Red. This replaces the mixed float/lateness rule. Zero float is not proof of a late finish. Keep float as supporting context, without adding a second colour trigger to this example. If you choose working days locally, replace the rule and formula together and document the calendar, holidays and counting convention.
For cost, use unrounded variance for classification and round only for display. Exactly 5% and exactly 10% are Amber; above 10% is Red. A valid underspend, zero variance and an overrun below 5% are Green on cost alone, unless the contingency Red condition applies. For example, 3% over budget with exhausted contingency and further overspend forecast is Red. A Green cost row does not demonstrate that scope or benefits are safe.
An unavailable, zero or negative approved budget makes this percentage rule unusable: cost is Not assessed unless an unresolved exception must be retained. Do not divide by zero or substitute zero variance. A missing, negative or unverifiable EAC is also unusable. Verify the agreed contingency treatment; if that Red trigger cannot be checked, flag the evidence gap rather than assuming it is false. An independently established Red condition still takes precedence.
Qualitative rows need the project’s agreed definitions of material, high, critical and appetite. Missing definitions or unresolved evidence contradictions prevent a complete assessment. Missing acceptance results are Not assessed, rather than a pass. A required check that is not yet due must be identified in the acceptance plan; it cannot be treated as a completed pass. If no checks are yet due, document that basis and any current acceptance concerns before assigning Green.
Risk exposure is separate from control completeness
An owner and a review date show that someone is managing a risk. They do not demonstrate low residual exposure. Under this starter rule, a high residual risk stays Amber even with a named owner, a review date and accepted exposure; a critical item stays Red. Reduce its colour only when reviewed evidence supports a lower exposure under the agreed scale. Acceptance records authority to carry exposure; it does not reduce that exposure by itself.
Show control completeness beside the exposure assessment: owner, response, next review date and overdue actions. Missing control fields trigger Review required for the report; they do not automatically change a known high exposure to Green or erase it into Not assessed. If residual exposure itself cannot be established, mark the current assessment Not assessed while retaining any open adverse exception and its last-supported severity.
The One-Page Calculator (Copy Into a Sheet)
Create one summary row per reporting cycle, backed by a dimension checklist and exception record. The following is a specification for a sheet, not a supplied Excel workbook. Keep formulas boring so anyone can audit them.
| Cell / column | What you enter or calculate |
|---|---|
| A — Cycle date | Reporting date and cut-off; criterion version and required dimension list in the supporting record |
| B — Baseline finish | Approved baseline end date and reference for the named finish or milestone |
| C — Forecast finish | Current forecast for that same finish or milestone; evidence reference, date and owner |
| D — Calendar days late | C-B after validating dates and removing times; no working-day claim |
| E — Time RAG | Green / Amber / Red / Not assessed using the time rule and exception precedence |
| F — Budget | Approved positive budget; reference and agreed contingency basis |
| G — EAC | Current verified non-negative estimate at completion on the same basis |
| H — Cost variance | (G-F)/F only after validation; retain full precision |
| I — Cost RAG | Cost bands plus the Red contingency trigger, evidence checks and retained exceptions |
| J–L — Scope / risk / benefits RAG | Rule-based dropdown: Green / Amber / Red / Not assessed; document approved exclusions separately |
| Additional dimension cells | Quality or other agreed required dimensions; all must feed the overall result |
| M — Computed RAG | Red first; then Not assessed for incomplete required assessment; otherwise worst colour, as specified below |
| N — Declared RAG | What you will put on the slide; defaults to M, including Not assessed |
| O — Override? | Yes / No |
| P — Override reason + approver | Mandatory if O = Yes; include approval reference, date and expiry cycle |
| Q — Road to green | One sentence + owner + date for non-green; evidence-gap action and deadline where applicable |
| Supporting record | Per-dimension source, evidence date, owner, age limit, freshness, confidence, control completeness, exceptions and closure requirements; overall completeness and report review state |
For usable inputs with no retained adverse exception, the starter classification is:
Time = IF(d > 15, "Red", IF(d > 5, "Amber", "Green"))
Cost = IF(v > 10% OR (contingency_exhausted AND EAC > budget), "Red",
IF(v >= 5%, "Amber", "Green"))
These are classification expressions, not full spreadsheet error handling. Validate evidence before applying them. Do not convert a blank, error or Not assessed value to zero. Encode Green = 0, Amber = 1, Red = 2 only for ordering usable colours; keep Not assessed as a separate state.
Missing evidence, open exceptions and overall precedence
Agree a maximum evidence age for each dimension. Measure age in calendar days from the evidence date to the cycle cut-off; evidence exactly at the age limit is within it. A source and observation must also be verifiable and unresolved contradictions must be reviewed. Freshness alone does not make evidence usable.
Use the same rules as the RAG status calculator companion:
- Preserve every unresolved adverse exception with its fact, last-supported severity, original evidence date and source, owner, closure requirement and current review state. Evidence ageing alone cannot close or downgrade it. A failed mandatory acceptance check stays Red until a passed retest or other authorised resolution satisfies the acceptance rules; record the closure evidence, date and reviewer. A colour override alone is not closure.
- Assess current evidence separately. A required dimension with missing, stale, contradictory or otherwise unusable evidence has a current assessment of Not assessed. Where an exception is retained, show both: for example, “Red — open failed acceptance check; current evidence stale”. For dimension roll-up, use the worse of the retained exception colour and usable current colour; if only one exists, retain that colour. If neither exists, the roll-up state is Not assessed.
- If any required dimension is Red, including a retained open Red exception, computed overall is Red. Keep every evidence gap visible and mark the assessment incomplete where required current evidence is unusable.
- If no dimension is Red but any required current assessment is Not assessed, computed overall is Not assessed. Retain known Amber facts beside it; do not replace the overall state with Amber.
- If all required current assessments are usable, take the worst dimension colour, including retained exceptions. Overall Green requires every required dimension to be Green and no unresolved Amber or Red exception. An empty or undefined required set is Not assessed.
Show completeness separately as Complete or Incomplete. Complete means every required current assessment is usable; it does not mean Green or that all exceptions are closed. Report review state is Ready for review when the assessment is complete, owners have confirmed it, confidence is adequate, required controls/actions are recorded and overrides are valid. Otherwise use Review required. A complete Red assessment can be Ready for review with the exception and response documented. Escalate urgent exceptions while the pack is being checked.
That sheet is the calculator for version one: transparent, portable and reviewable in a SteerCo pack. An interactive web widget can wait until the rules are stable.
Worked Example (Illustrative Numbers Only)
Suppose the forecast finish is 8 calendar days after the approved baseline finish, EAC is 7% over budget with contingency not exhausted, and one high residual risk has an owner and a review this week. Current evidence is usable. Scope is Green; benefits are excluded by prior agreement with a recorded rationale; no other dimensions are required in this example.
- Time → Amber (8 days is above 5 and no more than 15).
- Cost → Amber (7% sits in the 5–10% inclusive band).
- Risk exposure → Amber (the residual risk is high despite complete ownership and review fields).
- Computed RAG → Amber (all required dimensions assessed; worst colour Amber).
Declared RAG should stay Amber unless someone with authority overrides and logs why—for example, temporary cost pressure already funded by an approved change that the EAC field has not yet reflected. If you override to Green without fixing the inputs, you are doing watermelon reporting: green outside, red inside.
If the cost forecast were missing, overall would be Not assessed, with the known Amber schedule and risk facts still shown. If a required mandatory acceptance check had failed and remained open, overall would instead be Red even with that cost gap. Its evidence date becoming old would add a freshness warning, rather than remove the failure.
Override Log (Non-Negotiable)
Overrides happen. Sponsors ask. Politics exists. The failure mode is an override with no memory.
Minimum fields when Declared RAG differs from Computed RAG:
- Date and reporting cycle.
- Computed vs declared colour or assessment state.
- Reason in one sentence (fact, not hope).
- Approver name and role, approval evidence and approval date.
- Expiry: which future cycle the override dies unless renewed.
If you cannot name an approver, you do not have an override—you have a wrong status. An override cannot make missing evidence complete or close an unresolved adverse fact.
Road to Green
Every Amber or Red needs a recovery path that a stranger can understand in thirty seconds:
- Cause: which dimension tripped the rule.
- Action: what changes in the plan (not “monitor closely”).
- Owner: one name.
- Date: when the action lands, and when RAG should be re-computed.
- Ask: what decision you need from SteerCo, if any (money, scope, date, risk acceptance).
For Not assessed, name the missing evidence, who will supply or resolve it and by when. For an open acceptance failure, state the closure requirement as well as the next action. For messy recoveries, the practical reset moves in how to bring order to a messy project sit beside this RAG discipline—not instead of it.
How to Roll This Out in One Week
- Draft thresholds with the project board or PMO (use the starter table); agree required dimensions and evidence rules.
- Build the one-page sheet; lock column meanings and exception precedence.
- Run one cycle in parallel with your old slide colour—compare Computed vs Declared.
- Agree the override rule in SteerCo minutes.
- Replace unexplained colouring with the agreed evidence rules for this scorecard.
Checklist Before You Send the Pack
- Every required dimension has a stated input, complete rule and evidence reference/date.
- Green definitions and exact boundaries are agreed before formulas.
- Risk exposure is distinct from control completeness.
- Overall follows Red → missing required assessment → worst colour precedence; no averaging in this sheet.
- Open exceptions retain their facts, evidence dates and closure requirements even when evidence is stale.
- Completeness and report review state are shown separately from colour.
- Computed and declared results are both visible; overrides have approver and expiry.
- Non-green has cause, action, owner, date and ask; evidence gaps have an owner and deadline.
- Numbers match the schedule and cost files you would defend in the room.
FAQ
What does RAG mean in project management?
RAG means Red, Amber and Green: a summary of project health or delivery confidence against agreed expectations. Teams may use thresholds, evidence-informed professional judgement or both. InsightSpotter’s scorecard uses computable rules and a conservative worst-of roll-up; these are this method’s choices, rather than a universal definition of RAG.
How do you set RAG status criteria?
Agree the assessment purpose, material dimensions, meanings and evidence with the people using the report. For this scorecard, make Green, Amber and Red rules complete, specify exact boundaries, identify required dimensions and apply the stated exception and missing-evidence precedence before worst-of aggregation. Keep overrides and a road to green visible. As of October 2026, this remains an editorial method for local adaptation, not an industry scoring standard.
Should you average RAG colours across workstreams?
This scorecard uses worst-of for the agreed material dimensions and shows workstream RAGs separately where needed. Do not average its colour codes: that can hide an exception requiring a decision. Worst-of is the chosen roll-up here; it does not define every RAG scheme.
Further Reading
- How to Reduce Project Reporting Time
- What Happens When a Project Goes Over Budget
- What Is a Steering Committee (SteerCo)?
- What Are Project Constraints?
- AI in Project Management (2026) — where automation helps status drafting, and where human checks still own RAG
- AI Project Status Report Checklist — inputs, one bounded prompt and fact-checks before you circulate the pack
- RAG Status Calculator: Turn Evidence into a Reviewable Colour — evidence freshness, confidence and overrides on the same criteria
Want the next field note when it lands? Join the newsletter.

