Skip to content

03 · Earned Value Management Basics

Ask a project manager "are we on budget?" and the weak answer compares money spent against money planned. That comparison is close to meaningless: a team that has spent 60% of the budget might be 80% complete and thriving, or 30% complete and in serious trouble. Both look identical on a spend-versus-plan chart. Earned Value Management (EVM) fixes this by introducing a third number — the value of the work actually completed — so that cost performance and schedule performance can be separated and measured in the same units.

Every number in this module is computed from the same scenario, so you can verify each formula as it appears.

The three base measures

Everything in EVM derives from three quantities, all expressed in currency.

Term Full name Plain meaning
PV Planned Value Budgeted cost of the work scheduled to be done by now
EV Earned Value Budgeted cost of the work actually completed by now
AC Actual Cost What was actually spent to complete that work
BAC Budget at Completion The total approved budget for the whole project

The critical subtlety is that EV is measured in budgeted currency, not actual currency. If a task was budgeted at $10,000 and it is now finished, it has earned $10,000 of value — even if it cost $18,000 to complete. The $8,000 overrun shows up in AC, not in EV. Keeping those separate is what makes the whole method work.

EV requires a rule for crediting partially-complete work. Pick one and apply it consistently:

Rule How it credits Best for
0/100 Nothing until complete, then 100% Short tasks; most conservative
50/50 50% at start, 50% at completion Medium tasks, high volume
% complete Assessor's judgement of progress Long tasks — but easy to game
Milestone weighted Fixed credit at defined milestones Long tasks with verifiable checkpoints

The % complete rule is the most common and the most abused. "It's 90% done" for three consecutive months is the classic failure, and it corrupts every downstream number. Milestone-weighted credit is the usual cure.

The formula reference

Keep this table; it is the entire method on one page.

Metric Formula Reads as
Cost Variance CV = EV − AC Negative = over budget
Schedule Variance SV = EV − PV Negative = behind schedule
Cost Performance Index CPI = EV ÷ AC Below 1.0 = over budget
Schedule Performance Index SPI = EV ÷ PV Below 1.0 = behind schedule
EAC — current rate continues EAC = BAC ÷ CPI Default forecast
EAC — variance was one-off EAC = AC + (BAC − EV) Optimistic forecast
EAC — cost and schedule pressure EAC = AC + (BAC − EV) ÷ (CPI × SPI) Pessimistic forecast
Estimate to Complete ETC = EAC − AC Cash still required
Variance at Completion VAC = BAC − EAC Projected final over/underrun
To-Complete Performance Index TCPI = (BAC − EV) ÷ (BAC − AC) Efficiency now required

Two memory aids make these hard to get wrong. Everything starts with EV — every variance and index has EV as its first term. And variances subtract, indices divide; if the word is "variance" the answer is in currency, if it is "index" the answer is a ratio around 1.0.

A fully worked example

A 12-month office relocation programme. BAC = $480,000. At the end of month 4, the reporting pack shows:

Measure Value Where it comes from
PV $180,000 The baseline says $180,000 of work should be done by now
EV $150,000 Completed work, valued at its budgeted cost
AC $175,000 Invoices and timesheets actually booked

Step 1 — variances.

CV = EV − AC = 150,000 − 175,000 = −$25,000   (over budget)
SV = EV − PV = 150,000 − 180,000 = −$30,000   (behind schedule)

Step 2 — performance indices.

CPI = EV ÷ AC = 150,000 ÷ 175,000 = 0.857
SPI = EV ÷ PV = 150,000 ÷ 180,000 = 0.833

Read those out loud, because this is the sentence that goes in the status report: we are getting 85.7 cents of value for every dollar we spend, and we are progressing at 83.3% of the planned rate. Both are below 1.0, so the project is over budget and behind schedule simultaneously — the worst of the four quadrants.

Step 3 — sanity-check against the naive view.

View Calculation Result
Work complete 150,000 ÷ 480,000 31.25%
Budget spent 175,000 ÷ 480,000 36.46%
Work planned 180,000 ÷ 480,000 37.50%

A manager reporting only "we've spent 36% of budget in month 4 of 12" would sound comfortably ahead. In fact only 31.25% of the work is done. The gap between 36.46% and 31.25% is precisely the problem EVM exists to expose.

Step 4 — forecast the final cost. Three EAC formulas, three assumptions:

Assumption Formula Calculation EAC VAC
Overrun was a one-off AC + (BAC − EV) 175,000 + 330,000 $505,000 −$25,000
Current CPI continues BAC ÷ CPI 480,000 ÷ 0.857 $560,000 −$80,000
CPI and SPI pressure AC + (BAC−EV) ÷ (CPI×SPI) 175,000 + 330,000 ÷ 0.714 $637,000 −$157,000

The spread between $505,000 and $637,000 is the honest uncertainty, and the right move is to present all three with their assumptions rather than pick one. That said, the middle figure is the default for a reason: decades of programme data show that CPI tends to stabilise after roughly 20% of the work is complete and rarely recovers on its own. At 31% complete, the $560,000 forecast is the one to plan against. ETC = 560,000 − 175,000 = $385,000 is the cash still needed.

Step 5 — how good would we have to become?

TCPI = (BAC − EV) ÷ (BAC − AC)
     = (480,000 − 150,000) ÷ (480,000 − 175,000)
     = 330,000 ÷ 305,000
     = 1.082

To land on the original $480,000 budget, the remaining work must be delivered at 1.082 efficiency — by a team that has so far demonstrated 0.857. That is a required 26% improvement in cost efficiency, sustained for eight months, with no plan change. A TCPI more than about 5% above the demonstrated CPI is the signal to re-baseline rather than promise recovery. Reporting "we will catch up" here is not optimism, it is a forecast contradicted by your own data.

Step 6 — schedule forecast. If the SPI trend holds, 12 ÷ 0.833 = 14.4 months, so roughly 2.4 months late. Treat this as an indicator, not a schedule: EVM measures schedule in currency, so it cannot tell whether the slippage is on the critical path. Always confirm against the network (module 02) before quoting a date.

Reading the four quadrants

Two indices give four situations, each with a different response.

CPI SPI Situation Typical response
≥ 1.0 ≥ 1.0 Under budget, ahead Verify EV isn't inflated by a loose % complete rule
≥ 1.0 < 1.0 Under budget, behind Resources available — consider crashing
< 1.0 ≥ 1.0 Over budget, ahead Often bought speed with overtime; check it's deliberate
< 1.0 < 1.0 Over budget, behind Escalate; re-baseline is likely

The second row is the useful one. Being under budget and behind schedule usually means you are under-resourced, not efficient — and the money to fix it is already in the budget.

EVM is only as honest as the EV number

Every figure above descends from EV. If teams self-report percent complete against no verifiable criteria, EV inflates, CPI and SPI both look healthier than reality, and the EAC understates the overrun until it is too late to act. Tie EV credit to the same Definition of Done or milestone acceptance you use elsewhere. An EVM report built on optimistic self-assessment is worse than no report, because it carries false authority.

How It Actually Works

EVM's real power is that CPI and SPI are not independent status opinions, they're both ratios against the same planned-value curve, which is why they can and do move in opposite directions: SPI = EV/PV measures schedule efficiency (are you burning through planned scope on pace) while CPI = EV/AC measures cost efficiency (are you spending what that scope should cost) — a project can be running SPI = 1.10 (ahead of schedule) and CPI = 0.80 (over budget) simultaneously, which happens exactly when a team is throwing extra, expensive resources at the schedule to hit a date, buying schedule performance with cost performance. The EAC formula selection itself is a statistical judgment call, not a lookup: EAC = AC + (BAC−EV) assumes future work reverts to the planned rate (use only if the current variance was a one-off, non-recurring cause), while EAC = BAC/CPI assumes the current cost-performance trend continues unchanged for all remaining work (use when the cause is systemic — e.g., underpriced labor rates that won't improve) — picking the wrong one is choosing the wrong statistical model for how a variance propagates forward, not a matter of preference. TCPI, (BAC−EV)/(BAC−AC), is the algebraic answer to "what CPI must every future dollar achieve to still hit budget" — when TCPI exceeds roughly 1.10, it's a mathematical statement that recovery within budget requires efficiency the team has not yet demonstrated, i.e. a rebaseline is likely warranted, not a management failure.

Exercise

Set up EVM for a project of your own with at least eight work packages.

  1. Build the cost baseline: list each work package with its budgeted cost, and state your BAC. Spread the budget across at least six reporting periods to produce the PV curve.
  2. State which EV credit rule you are using for partially-complete work, and justify it in one sentence.
  3. Pick a status date roughly one third of the way through and record PV, EV and AC. Then compute, showing every step: CV, SV, CPI, SPI.
  4. Compute all three EAC variants, plus ETC and VAC for each. State which one you would report to the sponsor and why.
  5. Compute TCPI against BAC. Compare it to your demonstrated CPI and state plainly whether recovery to the original budget is credible — if it is not, say what you would re-baseline to.
  6. Write the three-sentence status paragraph you would actually send. It must quote CPI and SPI in plain language ("we are getting X cents of value per dollar"), give the forecast final cost, and name one corrective action.