Home / Blog / Project Controls

Project Controls

Building EVM Dashboards in Power BI That Leadership Reads

📅 9 Apr 2026 ⏱ 7 min read # Power BI · EVM

Earned Value Management has a perception problem. To a project director it can look like an acronym soup — PV, EV, AC, CV, SV, CPI, SPI, EAC, ETC, VAC — that costs effort to produce and delivers a table nobody reads. The fix isn't more metrics. It's a dashboard that answers the only two questions leadership actually asks: are we on track, and if not, where and how bad?

The three numbers everything else is built from

Strip EVM back and there are only three measured inputs. Get these right and the rest is arithmetic:

  • PV (Planned Value) — the budgeted cost of work scheduled to date.
  • EV (Earned Value) — the budgeted cost of work actually completed.
  • AC (Actual Cost) — what that completed work actually cost.

The discipline that makes or breaks a dashboard is how EV is measured. Subjective "percent complete" entered by hand will quietly corrupt every downstream metric. Wherever possible I tie EV to physical progress measurement rules — installed quantities, verified milestones — so the numbers are defensible in a progress meeting.

From inputs to insight

The derived metrics are simple, but each answers a distinct question:

SV  = EV − PV      → ahead or behind schedule (in £)
CV  = EV − AC      → under or over budget (in £)
SPI = EV / PV      → schedule efficiency (1.0 = on plan)
CPI = EV / AC      → cost efficiency (1.0 = on budget)

EAC = BAC / CPI    → forecast cost at completion
VAC = BAC − EAC    → forecast variance vs budget

SPI and CPI are the workhorses because they're unitless — a CPI of 0.92 means the same thing on a £2m package and a £200m programme, which makes them perfect for comparison across a portfolio.

The forecast is the deliverable Anyone can report what happened. Value comes from EAC and the TCPI — the cost performance you'd now need to still hit budget. When TCPI drifts far above your current CPI, you're telling leadership, with evidence, that the original target is no longer realistic. That's a conversation worth having early.

Designing the dashboard layer

On the Doha North Road project I built the reporting in Power BI on top of a clean data model. A few principles kept it usable:

  • One screen, top-down. Headline status (RAG on SPI/CPI) at the top, trend in the middle, detail on demand below. A director should get the story in five seconds and the cause in thirty.
  • Trends beat snapshots. A single CPI of 0.95 is ambiguous; a CPI sliding from 1.02 to 0.95 over three periods is a clear signal. Always plot the curve.
  • The S-curve stays central. Planned vs Earned vs Actual on one chart is still the most intuitive picture of project health ever invented. Everything else supports it.
  • Drill, don't clutter. Keep the top level clean and let users drill into WBS/CBS for the package dragging the numbers down.

Automate the boring half

EVM dies when it's a manual monthly grind. I push the repetitive work — pulling progress, reconciling actuals, recalculating indices — into a clean pipeline (Excel and Python for the heavy lifting, feeding Power BI). That does two things: it removes transcription errors, and it frees the analyst to do the part that needs judgement — explaining why the curve bent, not just that it did.

The honest caveat

EVM measures cost and schedule efficiency against a baseline. It says nothing about quality, safety, or whether the baseline was sensible to begin with. A green dashboard on a flawed plan is still a flawed plan. I always present EVM alongside the narrative — it's the evidence, not the verdict.

The takeaway

A good EVM dashboard isn't a wall of indices; it's a decision tool. Measure EV honestly, lean on SPI/CPI trends, make the forecast the headline, and automate the grind. Do that and project controls stops being a reporting tax and starts being the early-warning system it was always meant to be.