Project Health Checks, KPIs and Dashboards: Reporting That Drives Decisions
Every project sends reports, but not every report tells the truth, and fewer still lead to a decision. A weekly status email with every item marked green until the week everything turns red is the classic example. Good reporting is built on a small number of meaningful measures, clear definitions of what the colours mean and a dashboard designed for the person who will read it. This lesson covers project health checks, KPIs and dashboards that help sponsors act early.
Why Status Reports Fail
- Watermelon reporting: green on the outside, red inside. Teams avoid reporting problems, so everything looks fine until it is too late.
- Activity instead of outcome: “held 12 meetings” says nothing about progress.
- Too many metrics so nobody can see what matters.
- No definitions: one manager’s amber is another’s red.
- No decisions requested, so the report is read and filed.
The fix is to define a few indicators, agree thresholds in advance and make it safe to report bad news early. See Project Communication Management for the wider communication plan.
Leading vs Lagging Indicators
Lagging indicators report what has already happened: tasks completed, money spent, defects found. Leading indicators warn of trouble ahead: growing backlog of unresolved risks, rising scope change requests, key people at over 100 per cent allocation, test environments unavailable. A balanced dashboard needs both, with extra weight on the leading ones because they leave time to act.
Define RAG Status Clearly
Red, amber and green (RAG) status is only useful if everyone interprets it the same way. Write definitions in the plan, for example:
| Status | Schedule | Budget | Scope and quality | Risk and issues |
|---|---|---|---|---|
| Green | Within 5% of baseline | Within 5% of baseline | On track, no unresolved significant defects | No high risks without a response |
| Amber | 5% to 10% behind; recovery plan in place | 5% to 10% over; recovery plan in place | Minor scope or quality concerns being managed | High risks being actively managed |
| Red | More than 10% behind or a key milestone at risk | More than 10% over, or reserve exhausted | Material scope or quality problem | High risk or issue needing sponsor decision |
The thresholds above are illustrations. Set your own with the sponsor, and report overall status plus each dimension separately. Also state the trend (improving, stable, worsening) because a stable amber and a worsening amber need different responses.
A Practical KPI Set
| Area | Indicator | What it tells you |
|---|---|---|
| Schedule | Milestones met vs planned; schedule performance index (SPI) | Whether delivery is on track (see Earned Value Management) |
| Cost | Actual vs budget; cost performance index (CPI); forecast at completion | Whether money is being used as planned |
| Scope | Change requests raised and approved; requirements completed | Stability of scope |
| Quality | Defects open by severity; defect escape rate; test pass rate | Whether the product is fit for release |
| Risk | Number of open high risks; total risk exposure; overdue responses | Level of uncertainty |
| Issues and dependencies | Open issues, average age, blocked items | Whether the team is unblocked (see RAID Logs, Issues and Dependencies) |
| Team | Resource utilisation, turnover, pulse survey score | Sustainability and morale |
| Stakeholders | Sponsor satisfaction, decisions awaiting answer | Engagement and governance responsiveness |
| Benefits | Progress towards benefit targets | Whether the project is still worth doing |
For agile teams, add flow measures such as velocity trend, throughput and cycle time (see Scrum Metrics Explained). Choose about five to eight indicators, not all of them. Each should connect to a decision someone can make.
Designing a One-Page Dashboard
- Start with the audience. Executives need overall health, top three risks and decisions needed. Team leads need detail on defects, blockers and next steps.
- Put the overall status at the top, with trend arrows and a one-line summary.
- Show a few indicators visually: a milestone timeline, a budget burn line against plan, a risk heat map and a defect trend.
- Add three short lists: achievements this period, planned next period, and decisions or help needed.
- Keep it to one page and put detail in an appendix.
- Use consistent definitions and colours and avoid decoration that distracts from data.
Whatever tool you use, whether a spreadsheet, a project tool or a business intelligence dashboard, the principles are the same. Automate data collection where possible so the dashboard is not out of date by the time it is read.
Running a Project Health Check
A health check is a periodic, deeper review, perhaps monthly or at each phase gate, sometimes carried out by someone independent of the project team. Typical questions:
- Do all key stakeholders share the same understanding of objectives and scope?
- Is the plan current, and do the schedule, budget and resources still agree?
- Are the top risks known, owned and actively managed?
- Is the team stable, appropriately skilled and not overloaded?
- Is the sponsor engaged and are decisions made on time?
- Does the quality evidence, such as test results, support the planned release date?
- Is the business case still valid?
The output should be a short list of findings with severity, recommended actions and owners. If the findings point to serious trouble, see Recovering a Troubled Project.
Worked Example: Data Platform Project
A data platform project has reported green for four months. The project manager introduces a one-page dashboard with agreed thresholds. The first report shows schedule green but two leading indicators amber: the number of open high risks has doubled and two data engineers are allocated at 130 per cent. The sponsor asks for details, approves a short extension of the contractor and delays a lower-priority feature. Two months later the project still meets its main milestone, because the leading indicators prompted action before the schedule itself turned red.
Common Mistakes
- Reporting only what has been completed, not what is likely to happen.
- Changing the thresholds when results look bad.
- Putting twenty metrics on the page and no decisions.
- Punishing the messenger, which teaches the team to hide problems.
- Producing a polished report nobody reads rather than a short one that triggers action.
- Define RAG status with measurable thresholds and report trend as well as colour.
- Use a handful of indicators, mixing leading and lagging measures.
- Design the dashboard for its audience and always include decisions needed.
- Run periodic health checks and make it safe to report bad news early.