PMO dashboard: 12 metrics to report, with definitions

Every Monday you pull 12 project trackers into one portfolio view. Most projects are green, 2 managers use their own idea of amber, and 1 project hasn't had an update since September. Then the steering committee asks why the data migration is amber, and the meeting goes to defending a color.

The fix is a short list of metrics, each with one definition, one formula, and one source. Here are 12 metrics a project or program office can report, with earned value formulas from US government guidance and a way to stop status colors from drifting.

What should a PMO dashboard answer?

A portfolio dashboard serves one decision: which projects need help this week, and what kind. If a number never changes what the steering committee does, it belongs in a project report.

That leaves 4 questions. Is the work on schedule? Is it on budget? What could go wrong, or already has? Can we trust the status? See a smaller set on one page in the PMO portfolio example.

Which 12 metrics do project offices report?

The earned value terms are explained in the next section.

How do you calculate schedule and cost metrics with earned value?

The US Department of Energy's EVMS Gold Card defines the inputs: planned value (PV) is the budgeted cost of work scheduled, earned value (EV) is the budgeted cost of work performed, and actual cost (AC) is the actual cost of work performed. Budget at completion (BAC) is the total budget. GAO's Cost Estimating and Assessment Guide reads the indexes plainly: a CPI below 1.0 means the work cost more than planned, and an SPI below 1.0 means less work was done than planned.

Say a project has a $400,000 budget. By the status date it planned $200,000 of work (PV), finished work budgeted at $160,000 (EV), and spent $190,000 (AC).

• Schedule. SV is -$40,000 and SPI is 0.80: 80 cents of work done per dollar scheduled.
• Cost. CV is -$30,000 and CPI is 0.842: each dollar spent bought 84 cents of work.
• Forecast. EAC is $400,000 ÷ 0.842, about $475,000, or $75,000 over budget.
• Budget used. $190,000 ÷ $400,000 is 47.5%, which looks fine on its own.

So budget used should never sit alone: it compares spending with the budget, not with the work done. GAO makes the same point: when percent spent runs well ahead of percent complete (47.5% against 40% here), the project is overrunning.

GAO adds 2 cautions. Schedule variance fades by itself: once all the work is done, EV equals PV, so even a late project ends with an SPI of 1.0. GAO notes that SPI "loses its predictive value as a program progresses," so late in a project, watch milestone dates instead. And test the manager's own forecast: if the to-complete performance index (TCPI, the efficiency the remaining work would need) is more than 5 percentage points above the CPI, GAO's rule of thumb calls the forecast too optimistic. If the project above still forecasts $400,000, its remaining work, budgeted at $240,000, must be done for $210,000: a TCPI of 1.14 against a CPI of 0.842.

Earned value needs a budget per piece of work and an honest percent complete. If your projects lack either, drop SV, SPI, CV, and CPI, and show the manager's estimate to complete, labeled as an estimate.

How do you count milestones, risks, issues, and changes?

Milestones hit on time is milestones done by their due date divided by milestones due so far. GAO's Schedule Assessment Guide describes a looser cousin, the baseline execution index: activities completed divided by activities that should have been completed by the status date. It counts a late finish as a finish. Pick one and write it down. Say 31 of 36 milestones due this quarter were on time: 86%, with 3 undated milestones left out and listed by name.

Milestones slipped counts milestones past due and not done. Show days late, because 1 milestone 40 days late matters more than 3 that are 2 days late.

Risks and issues differ. The UK government's common glossary defines an issue as "a relevant event that has happened, was not planned and requires management action," and a risk as "the effect of uncertainty on objectives." Report open risks by the rating scale in your register, and open issues with their age. An issue open for 60 days is a decision nobody has made.

Pending change requests counts requests submitted but not yet decided, with their total cost and schedule impact.

Resource utilization is hours logged divided by hours available. Take holidays out of hours available, or a team on vacation looks underused. Above 100% means someone is overbooked.

Missing status updates counts projects with no update in 7 days. GAO's cost guide warns that when open work isn't statused regularly, the schedule may not really be used to manage the program. Show a missing update as missing, not as last week's color.

How does RAG status get gamed, and how do you fix it?

Red, amber, green (RAG) status is the easiest number on a PMO dashboard to bend.

Why do status colors drift?

Nobody has to lie for the color to drift:

• The person being judged sets the color. Writing about risk interviews, GAO's schedule guide calls this motivational bias: people shade answers when they fear punishment from someone in authority.
• Amber has no definition. If amber means "some concerns," a project can stay amber while a milestone slips 6 weeks.
• Bad news travels slowly. A 2014 MIT Sloan Management Review summary of 14 studies on IT and software projects lists 5 "inconvenient truths," including these 2: executives can't rely on staff to report status accurately, and they often ignore bad news.
• Stale colors carry forward. Without a date, last month's green looks like this week's.
• Roll-ups hide the red. GAO makes this point about cost forecasts: build them below the program level so weak areas aren't masked by strong ones.

How does a written rule fix it?

Make the color a calculation: write the rule once, compute it from the data, and show it next to the color. An example to adapt, not a standard:

Then let people disagree in the open. The project manager can report a different color, and the dashboard shows both: "Amber by rule (milestone 6 days late). PM reports green." The debate moves from whose opinion wins to whether the plan or the rule should change.

UK government major projects use a similar check. Each senior responsible owner reports a delivery confidence rating every quarter, but where the National Infrastructure and Service Transformation Authority (NISTA) has supported or independently assured a project in the past 6 months, its own rating is published instead. Its amber has a written meaning: delivery appears feasible, but significant issues need management attention.

Keep thresholds fixed for a quarter and date any change, or this quarter's 9 green projects can't be compared with last quarter's 7.

Where does the data for each metric usually live?

PMO numbers usually come from 6 places: a weekly status sheet, project plans exported from a scheduling tool, a finance export of actual costs, a RAID log (risks, assumptions, issues, dependencies), a change log, and timesheets. Each has its own cutoff: finance may close on the 5th working day, status every Friday.

Put each source's date on the dashboard: "Budget used 61% (costs through September 30, status through October 9)." See how to show data freshness. Writing each definition once in a metrics dictionary stops managers from quietly using their own.

Where Every Number fits

Every Number builds dashboards where every number shows its work: definition, filters, rows used, rows left out and why, source file, and the date the data runs through. Today that is a $99 concierge pilot. You send one spreadsheet, such as your portfolio tracker, and a person builds 5 to 7 metrics in 5 business days. You approve each definition, including your RAG rule, before anything is built. See the PMO use case or the pilot.

It is not a good fit if a portfolio tool your steering committee trusts already reports earned value, or if every project keeps its own file and you can't combine them into one spreadsheet. And 5 to 7 is not 12, so start with the metrics your committee decides with.