The one test every dashboard tile must pass
Before a metric earns a place on the page, it has to answer: "if this number moves, does someone do something different?" If the answer is no, it's decoration — cut it. A task-count that never changes a decision is noise dressed up as rigour. The best dashboards are ruthless about this; they'd rather show six numbers that matter than forty that don't.
The five things worth tracking
Across almost every project, decision-driving information falls into five lanes:
| Lane | The question it answers | Lead metric |
|---|---|---|
| Schedule | Will we finish on time? | SPI, milestone variance, forecast finish |
| Cost | Will we finish on budget? | CPI, forecast at completion (EAC) |
| Scope | Are we still building the agreed thing? | change requests, scope added vs baseline |
| Risk | What could still go wrong? | top exposures, trend, P80 finish |
| Flow (agile) | Is delivery steady? | throughput, cycle time, WIP |
Two of these — cost and schedule — are best expressed through earned value, because CPI and SPI put time and money on the same comparable scale. Risk is best expressed as a forecast range (P50/P80), not a single date.
RAG status, done honestly
Red/Amber/Green is the most abused device in project reporting. "Green" too often means "I don't want a hard conversation this week." A defensible RAG has rules, not moods:
- Green — forecast within tolerance on time and cost; no unmitigated high risks.
- Amber — forecast breaching a threshold, with a recovery plan in hand.
- Red — forecast breaching, no agreed recovery, or a high risk realised.
Write the thresholds down once and apply them mechanically. A status that anyone can audit against the numbers is worth ten that depend on the reporter's optimism.
The watermelon problem: green on the outside, red in the middle. It happens whenever status is a feeling instead of a calculation. Tie every colour to a number on the same page and the watermelon can't form.
Lead vs lag: show the windscreen, not the mirror
"Tasks completed this month" is a lag indicator — it tells you about the past. "Forecast finish date" and "CPI trend" are lead indicators — they tell you where you're heading. A dashboard built only from lag metrics is a rear-view mirror; you'll see the wall after you hit it. Weight the page toward forecasts and trends.
One page, one source of truth
If schedule lives in one tool, cost in a spreadsheet, and risk in someone's head, your "dashboard" is really three arguments waiting to happen. The value of a single control sheet is that every tile is computed from the same underlying data, so the numbers can't quietly disagree. When the board asks how a figure was derived, you open it — you don't go hunting.
Keep it defensible
The final rule ties the rest together: every number on the dashboard should be traceable to its inputs. A RAG you can't explain, a percentage nobody can reproduce, a forecast with no method behind it — these are exactly the figures that collapse under one sharp question in a steering committee. Build the page so that opening any tile shows the working.
A one-page control system that computes every tile
The Day-One Toolkit is a 40-sheet Excel control system — schedule, earned value, 5×5 risk and agile flow — that rolls up to a single board-ready dashboard, every formula visible and offline. Start with the free Field Guide.