Skip to content

09 · Reporting & Dashboards

Most project reports fail in the same way: they describe activity rather than answer a question. Three pages of "the team continued integration work" tells a sponsor nothing they can act on. A report exists to support a decision — if nothing in it could change what anyone does, it should not have been written. This module covers how to design a report around its audience, which metrics belong on a dashboard, how to set red/amber/green thresholds that mean something, and how to read a milestone trend that is quietly telling you the forecast is fiction.

Start with the decision, not the data

Before designing anything, answer three questions for each audience: what decisions do they make, at what frequency, and what is the smallest set of numbers that lets them make those decisions well?

Audience Decision they make Cadence Depth Format
Delivery team What do I work on next; what is blocked Daily Task level Board + stand-up
Project manager Where to intervene this week Weekly Work package Full dashboard
Sponsor Approve changes, remove obstacles, fund Fortnightly Milestone + EVM 1 page + exceptions
Steering committee Continue / adjust / stop; approve major change Monthly Objectives + risk 1 page + decision log
Portfolio office Compare against other projects Monthly Standard indices only Automated feed

The mistake is producing one report and sending it to everyone. The team's board is meaningless to a steering committee, and the steering pack is far too coarse for the team. Same underlying data, different aggregation.

What belongs on a one-page dashboard

Nine panels are enough. More than that and the reader stops scanning and starts skipping.

Panel Content Answers
Overall RAG One letter, plus a one-line reason Is this project in trouble?
Schedule SPI, next milestone, forecast vs baseline date Will it land on time?
Cost CPI, AC vs BAC, EAC and VAC Will it land on budget?
Scope Stories/deliverables done vs total; approved changes Are we still building the same thing?
Quality Open defects by severity, escaped defects Will it actually work?
Top risks Top 5 by exposure, with owner and next action What could still go wrong?
Issues Open issues past due, with owner What has already gone wrong?
Decisions needed Decision, options, cost of delay, needed-by date What do you need from me?
Milestone trend Forecast date for one key milestone, period by period Is the forecast credible?

The Decisions needed panel is the one that makes sponsors read the report. If every report you send asks for nothing, you have trained your audience that your reports are optional.

The status report template

Field Content
Project / period Warehouse Management System rollout — Period 8, w/e 24 Oct
Overall status Amber — schedule slipping; cost favourable; one decision required
Status last period Amber (2nd consecutive period)
Achieved this period Site 3 cutover complete; integration test cycle 2 passed
Planned next period Site 4 cutover; UAT entry review; performance retest
Schedule SPI 0.92; UAT complete forecast 21 Oct vs baseline 30 Sep (15 working days late)
Cost CPI 1.15; AC $400,000 of BAC $1,000,000; EAC $869,565 (VAC +$130,435)
Scope 46% complete by value; 3 changes approved, 1 rejected this period
Quality 18 open defects (2 severity-1); 9 escaped from the last release
Top risk Site 4 network upgrade may slip — owner: Infrastructure Lead
Issue past due Master data cleansing 6 days late — owner: Ops Manager
Decision required Approve 2 extra testers ($34,000) by 31 Oct or UAT slips a further 10 days

Note that the report is honest about the second consecutive amber and gives the decision a deadline with a stated consequence. "Please advise" is not a decision request; "approve by 31 Oct or lose 10 days" is.

Worked example — the EVM panel

Everything in the cost and schedule rows above comes from four numbers taken at the end of Period 8.

Measure Value
BAC — Budget at Completion $1,000,000
PV — Planned Value $500,000
EV — Earned Value $460,000
AC — Actual Cost $400,000

Everything else is derived:

Derived metric Formula Calculation Result
Cost Variance EV − AC 460,000 − 400,000 +$60,000
Schedule Variance EV − PV 460,000 − 500,000 −$40,000
CPI EV ÷ AC 460,000 ÷ 400,000 1.15
SPI EV ÷ PV 460,000 ÷ 500,000 0.92
EAC (typical variance) BAC ÷ CPI 1,000,000 ÷ 1.15 $869,565
ETC EAC − AC 869,565 − 400,000 $469,565
VAC BAC − EAC 1,000,000 − 869,565 +$130,435
TCPI (to BAC) (BAC − EV) ÷ (BAC − AC) 540,000 ÷ 600,000 0.90

Three ratios also belong on the panel because together they tell the story at a glance:

  • Percent complete = EV ÷ BAC = 46.0%
  • Percent spent = AC ÷ BAC = 40.0%
  • Percent scheduled = PV ÷ BAC = 50.0%

Read them in that order: we planned to be 50% through, we are 46% through, and we have spent 40%. The project is behind schedule and getting good value for money. TCPI of 0.90 says the remaining work may be delivered at 90% of the budgeted rate and still hit BAC — there is genuine cost headroom, which is exactly why the $34,000 tester request is affordable.

Distrust a very favourable CPI early on

A CPI well above 1.0 in the first third of a project is more often an artefact than an achievement. The usual causes are supplier invoices that have not yet been accrued, timesheets booked to the wrong code, or work being claimed complete more generously than the completion rules allow. Before reporting CPI 1.15 as good news, check accruals and check how percent-complete was claimed. Report the number, but report the caveat with it.

RAG thresholds that mean something

Most RAG statuses are a feeling. Publish the rules instead, agree them with the sponsor at kickoff, and apply them mechanically.

Dimension Green Amber Red
Schedule (SPI) ≥ 0.95 0.90 – 0.94 < 0.90
Cost (CPI) ≥ 0.95 0.90 – 0.94 < 0.90
Key milestone On or ahead of baseline 1–10 working days late > 10 working days late
Open severity-1 defects 0 1–2 3 or more
Risk exposure vs reserve < 60% of reserve 60–90% > 90%
Overall All green Any amber Any red, or amber twice running with no recovery plan

Applying the rules to the worked example: SPI 0.92 → Amber. CPI 1.15 → Green. Key milestone 15 working days late → Red. Two severity-1 defects → Amber. Overall status is therefore Red by the milestone rule, even though the report above says Amber — which is precisely the argument the PM must have with themselves before publishing. If your rules say red, report red.

Watermelon reporting

Green on the outside, red on the inside. It happens when status is set by judgement rather than rule, and it always ends the same way: a project that was green for eleven months goes red in month twelve and the sponsor's first question is why nobody told them. Mechanical thresholds exist to protect you from your own optimism — and from the pressure to be green that every reporting culture applies.

Milestone trend analysis

The single most revealing chart on any dashboard is not a burndown. It is the forecast date for one key milestone, recorded once per reporting period.

Period reported Forecast for "UAT complete" Slip vs baseline (working days) Slip added this period
P4 30 Sep 0
P5 30 Sep 0 0
P6 7 Oct +5 +5
P7 14 Oct +10 +5
P8 21 Oct +15 +5

This table says something no single-period status can: the milestone is slipping one week per week, and has done so three periods running. A forecast that moves by exactly one period every period is not a forecast at all — it is the team reporting that they have not started the recovery. Extrapolating the trend, the milestone will not be met until the underlying cause is addressed, regardless of what date is currently written down.

Ask two questions whenever you see this shape: what changed in P6 to start the slide, and what specific action is meant to stop it? "The team will work harder" is not an answer; "two testers approved by 31 Oct" is.

Leading and lagging indicators

Most dashboards are entirely lagging — they measure what has already happened. Add at least two leading indicators, which move before the outcome does.

Type Metric What it tells you
Lagging SPI, CPI, milestones met, defects escaped Where you ended up
Leading Requirements still unapproved Future rework and change requests
Leading Average age of open blockers Whether impediment removal is working
Leading Test cases written vs stories accepted Whether QA will become the bottleneck
Leading Risks with no owner or no next action Whether risk management is real or ceremonial
Leading Unplanned work as % of capacity Whether the plan is being overrun by interrupts

A rising average blocker age is visible weeks before it shows up as a fallen SPI. That gap is the entire value of a leading indicator.

Reporting anti-patterns

Anti-pattern Why it fails Fix
Percent complete by task count 90 of 100 tasks done says nothing about remaining effort Weight by value or effort; use EV
Narrative with no numbers Cannot be compared period to period Every claim gets a metric
Numbers with no narrative Reader cannot tell cause from noise One sentence of why per red or amber
Report built by hand each period Consumes a day; drifts in definition Automate extraction; hand-write only commentary
Metrics used to judge people People manage the metric, not the work Report at project level, never individual level
Everything green until it isn't Judgement-based status Published thresholds, applied mechanically

The last row of that table is the one worth writing on the wall. Define each metric in writing once — what counts as "done", which cost codes feed AC, how percent-complete is claimed — and never change the definition mid-project without saying so on the report. A metric whose definition moves silently is worse than no metric, because it looks comparable when it is not.

How It Actually Works

RAG thresholds that "mean something" are statistical control limits applied to project metrics the same way Module 6's control charts apply them to defect data: a threshold set at CPI/SPI < 0.90 → Red isn't an arbitrary round number, it should be calibrated against the historical variance of completed projects of similar size in your organization — a threshold too tight (Red at 0.98) generates false alarms and trains stakeholders to ignore the dashboard; a threshold too loose (Red at 0.70) means real trouble hides in Amber for months. Milestone trend analysis works because a milestone date, re-forecast at every reporting period, produces a time series whose slope is diagnostic: a flat line (each report re-forecasts the same date) signals a stable plan, a line that consistently drifts later by roughly the same amount each period signals a systemic, recurring cause (not a one-off), and a line drifting later by an accelerating amount is the leading indicator of exactly the kind of variance that triggers a Level 3 Module 8 recovery review — long before EVM's own lagging cost/schedule indices would show the same trouble, because milestone-date slippage is a leading indicator while CPI/SPI are lagging indicators computed from work already done.

Exercise

Design the reporting for a real project.

  1. Build the audience table for your project with at least four audiences, each with the specific decision they make, the cadence, and the format. Identify one report you currently produce that no listed audience needs.
  2. Take a real reporting period and compute the full EVM panel: CV, SV, CPI, SPI, EAC, ETC, VAC and TCPI, showing the arithmetic. Then write the three ratios (percent complete, spent, scheduled) and the one sentence you would put on the dashboard to explain them.
  3. Publish your RAG thresholds as a table covering at least four dimensions, with a rule for overall status. Apply them mechanically to your current period and state whether the honest colour differs from the one you would have chosen by instinct.
  4. Build a milestone trend table for one key milestone across at least five past periods. State the per-period slip rate and what it predicts.
  5. Choose three leading indicators for your project and explain, for each, how many weeks ahead of SPI or CPI you expect it to move.
  6. Write one complete status report against the template. Include a Decision required row with the options, the cost of each, and the date beyond which the decision costs schedule.