How to Set Up a Project Health Score System

Most project boards are flying blind. By the time a monthly report lands, its data is three-to-six weeks old — a snapshot of a project that has already moved. A project health score fixes this by rolling the signals that actually predict trouble into a single, live indicator every stakeholder can read at a glance. Done well, it turns "how's the project doing?" from a two-day data-gathering exercise into a number on a dashboard.

This guide covers what to measure, how to weight it, and how to keep it honest.

What a health score is (and isn't)

A health score is a weighted composite of a handful of objective KPIs, normalised to a common scale (say 0–100 or Red/Amber/Green). It is not a vanity metric, and it is not a replacement for judgement. Its job is to concentrate attention: green means "no action needed," amber means "look here this week," red means "escalate now."

The failure mode to avoid is a score built from whatever data is easy to collect. If it doesn't move before a problem becomes visible, it's decoration.

The five dimensions that actually predict trouble

For construction and real-estate delivery, five dimensions carry most of the signal:

• Schedule. Are we ahead of or behind the baseline? Use the Schedule Performance Index (SPI = earned value / planned value) or a duration-weighted progress-vs-plan variance. A simple "% complete" is not enough; it hides whether the right work is done.
• Cost. Cost Performance Index (CPI = earned value / actual cost), plus committed-vs-budget. A project can look on-track on cash but be over-committed.
• Quality. Open NCRs, ageing of non-conformances, and inspection pass rate. Rising open NCRs is an early quality signal.
• Safety (HSE). Incident frequency, open high-potential hazards, overdue safety actions. A safety trend is often the first sign of a programme under pressure.
• Risk & compliance. Open high-score risks, overdue statutory items, expiring insurances/permits, and overdue approvals.

Weighting: make the trade-offs explicit

The weights are your priorities, so set them deliberately. A defensible starting point:

• Non-negotiables can veto. A serious safety event or a lapsed statutory permit should force the whole project to red regardless of the composite — some things aren't tradeable against schedule.
• Weight by phase. Procurement-heavy early phases should lean on cost and long-lead risk; fit-out and handover phases should lean on quality and snagging.

Normalise before you combine

Each KPI lives on a different scale, so normalise to 0–100 before weighting. For index-style metrics, a common convention is: CPI/SPI ≥ 1.0 → 100 (on or ahead); 0.95–1.0 → amber band (≈ 70–100); < 0.95 → scales down toward 0.

Pick your "on track" threshold and apply it everywhere — inconsistent thresholds across metrics are the most common reason a score loses trust.

Automate it — or it won't survive contact with reality

A health score that someone assembles by hand every month will drift, then quietly die. The whole point is that it updates itself from data you're already capturing:

• Schedule and cost pull from your programme and earned-value baseline.
• Quality and safety pull from your NCR, inspection and incident registers.
• Risk and compliance pull from your risk register and statutory tracker.

Rolling up to a portfolio

For a PMC running many projects, the same discipline rolls up: a portfolio view of every project's health lets a director triage — spend the week on the two red projects, not the twelve green ones. The key is that each project's score is computed the same way, so they're comparable.

Common mistakes

• Too many inputs. Five to seven KPIs is plenty. Twenty makes the score inscrutable and un-actionable.
• Lagging data. If the inputs are monthly, the score is monthly. Feed it from live registers.
• Gaming. If teams can move the number without improving the project (e.g. closing NCRs without fixing the cause), the metric is corrupt. Tie inputs to objective evidence.
• No thresholds for action. A number with no "what do I do at amber?" is trivia. Define the response for each band.

The payoff

A well-built health score changes the meeting. Instead of an hour reconstructing what happened, the board spends the hour on the two things that are amber. That's the entire value: attention, directed to where it's needed, from data that's already true.