Earned value management (EVM) remains the most reliable way to know, at a given date, whether a project is genuinely ahead or behind, under or over budget. Its second edition has just been republished by ISO. The method itself isn’t the problem — it’s that it’s rarely taken past a simple percent-complete figure. This piece works through the formulas, then applies them to a full worked case.
What ISO 21508:2026 changes
ISO/TC 258, which PMI administers on ANSI’s behalf, published the second edition of ISO 21508 in 2026, replacing the 2018 edition. The main change: risk management is now integrated into the earned value management process steps, with added clauses on integrating EVM with other control processes. The standard remains a practice guide — generic, applicable to any type of organisation and project — not a certifying framework. It complements ISO 21512:2024, which covers operational implementation, and PMI’s standard, which is more prescriptive.
The three base measures
EVM rests on three figures, all measured at the same status date and expressed in the same currency:
- PV (Planned Value): the value of work that was supposed to be done by this date, per the baseline plan.
- EV (Earned Value): the value of work actually completed by this date, valued at the budgeted rate — not the actual cost.
- AC (Actual Cost): what has actually been spent to produce that work.
The most common confusion is around EV: it’s neither a self-reported percent-complete figure nor the amount spent. It’s the budgeted value of work genuinely delivered, verifiable against milestones or deliverables.
The indices that drive the decision
| Indicator | Formula | Reading |
|---|---|---|
| CV — Cost Variance | EV − AC | Negative: delivered work cost more than planned |
| SV — Schedule Variance | EV − PV | Negative: the project delivered less value than planned by this date |
| CPI — Cost Performance Index | EV / AC | Below 1: every dollar spent produces less than a dollar of value |
| SPI — Schedule Performance Index | EV / PV | Below 1: value is being produced more slowly than planned |
Case study: migrating a lending module to a new core banking platform
A fictional but representative project: migrating the lending module of a mid-size bank to a new core banking platform. Total budget (BAC, Budget at Completion): $2,400,000. Planned duration: twelve months, split into five work packages.
- Package 1 — Scoping and target architecture: complete, on budget.
- Package 2 — Interface development: in progress, behind schedule, because a third-party API delivery ran six weeks late.
- Package 3 — Data migration: in progress, over budget, because legacy data required cleansing work that wasn’t costed at the scoping stage.
- Package 4 — Integration testing: not started.
- Package 5 — Cutover and hypercare: not started.
Status date: end of month 6, at the planned midpoint.
| Measure | Value | Source |
|---|---|---|
| BAC | $2,400,000 | Approved total budget |
| PV | $1,320,000 | 55% of budget, per the baseline plan |
| EV | $1,080,000 | 45% of budgeted value actually delivered |
| AC | $1,250,000 | Actual spend to the same date |
Variances and indices:
- CV = EV − AC = 1,080,000 − 1,250,000 = −$170,000
- SV = EV − PV = 1,080,000 − 1,320,000 = −$240,000
- CPI = EV / AC = 1,080,000 / 1,250,000 = 0.86
- SPI = EV / PV = 1,080,000 / 1,320,000 = 0.82
The diagnosis is clear: the project is both behind schedule (SPI 0.82 — only 82% of planned value delivered) and over budget (CPI 0.86 — every dollar spent produces only $0.86 of value). The data migration package accounts for most of the CV; the interfaces package accounts for most of the SV.
Three ways to forecast the finish
This is where most project tracking stops too early — at the variances, without a forecast. EAC (Estimate at Completion) answers an operational question: given what we’re observing today, what will this project actually cost? There are several formulas, depending on the assumption made about the remaining work.
| Method | Formula | Assumption | EAC | VAC (BAC − EAC) |
|---|---|---|---|---|
| Isolated variance | AC + (BAC − EV) | The overrun observed was a one-off; remaining work proceeds at the planned rate | $2,570,000 | −$170,000 |
| CPI holds | BAC / CPI | Current cost performance continues unchanged through completion | $2,778,000 | −$378,000 |
| CPI × SPI | AC + (BAC − EV) / (CPI × SPI) | The schedule slip also drives up the cost of remaining work (overtime, added resources, knock-on effects) | $3,117,000 | −$717,000 |
The spread between the three methods isn’t cosmetic. The first, most optimistic method assumes an isolated overrun and points to a contained $170,000 overrun. The third — which accounts for the fact that the interface delay will likely also push up the cost of data migration and testing — puts that figure at more than four times as much: $717,000, close to 30% of the original budget. On this project, where the delay (package 2) and the cost overrun (package 3) sit on interdependent packages — integration testing in package 4 can’t start until packages 2 and 3 are stable — the third method is the realistic one.
One more indicator confirms the diagnosis: TCPI (To-Complete Performance Index), the performance the remaining work would need to hit to still meet the original budget. Here, TCPI = (BAC − EV) / (BAC − AC) = 1,320,000 / 1,150,000 = 1.15. The team would need to produce $1.15 of value for every dollar spent on the rest of the project — against the $0.86 it’s currently producing. The gap between the two (1.15 versus 0.86) signals that recovering to the original budget is unrealistic without a change in scope, resourcing, or method.
What this means for you
- Never calculate EAC with a single method. The isolated-variance method consistently understates risk when SPI is also degraded. Cross-check at least the CPI-holds method against the CPI × SPI method, and use the more conservative one whenever the packages that are late and the packages that are over budget depend on each other.
- Set an alert threshold on CPI and SPI, not just on percent complete. A common threshold is 0.90: below it, trigger a formal governance review rather than waiting for the next scheduled steering committee.
- Recalculate TCPI at every status point. A TCPI persistently above 1.10–1.15 across several consecutive measurements signals that the original budget target is no longer achievable — that’s the moment to formally rebaseline the BAC rather than issue another unrealistic recovery plan.
Building a reliable EVM dashboard — with regular status points and governance that reacts to the indices rather than the schedule alone — is a skill that’s learned and practised. Helios Advisory designs training pathways in project management and commercial development for PMO teams running this kind of programme.
Leave a Reply