Cumulative CPI is the Cost Performance Index calculated from day one of the project to the current reporting date. It divides total earned value by total actual cost across the entire project life to date. Unlike period CPI, which bounces around from month to month, cumulative CPI stabilises over time and becomes the most reliable predictor of where your final cost will land.
Cumulative CPI = Sum of all EV to date / Sum of all AC to date
Or simply: Cumulative CPI = EV(cumulative) / AC(cumulative)
This is the CPI that feeds into EAC formulas, drives trend-based forecasting, and, critically, almost never recovers once it drops below 0.90 past the 20% completion mark. That last point isn't theory. The US DoD proved it across hundreds of contracts, and I've seen it hold on NEC4 projects in the UK.
Cumulative CPI is part of the earned value definitions glossary. For the full CPI and SPI breakdown, see the CPI and SPI formulas page.
Cumulative vs Period CPI: The Trend vs the Snapshot
The difference matters more than most teams realise.
CUMULATIVE CPI vs PERIOD CPI – 12-Month Tracking Month Period CPI Cumulative CPI ────── ────────── ────────────── 1 1.12 1.12 ← Early data, both volatile 2 0.84 0.97 ← Period tanks (big invoice), cumulative absorbs it 3 0.95 0.96 ← Period recovers 4 1.05 0.98 ← Good month 5 0.78 0.94 ← Period CPI crash (subcontractor spike) 6 1.02 0.95 ← Period looks fine, cumulative barely moves 7 0.88 0.94 ← Another bad month 8 0.91 0.93 ← Cumulative settling around 0.93 9 1.05 0.94 ← Good period, cumulative ticks up slightly 10 0.89 0.93 ← Bad period, cumulative barely flinches 11 0.93 0.93 ← Convergence – both telling the same story 12 0.90 0.93 ← Cumulative locked in. This IS the project's efficiency. Period CPI Cumulative CPI 1.15 ┤ 1.15 ┤ │ * │ 1.05 ┤ * * 1.05 ┤ │ │ * 0.95 ┤ * * * * 0.95 ┤ * * * * │ │ * * * * * 0.85 ┤ * * * * 0.85 ┤ │ │ 0.75 ┤ * 0.75 ┤ └──┬──┬──┬──┬──┬──┬──┬──┬──┬── └──┬──┬──┬──┬──┬──┬──┬──┬──┬── 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 VOLATILE – reacts to every STABLE – shows the real trend invoice and progress swing after ~3-4 months of data
Period CPI is noise with signal buried inside it. Cumulative CPI is signal with the noise averaged out. That's why every serious EAC formula uses cumulative CPI as its primary input.
Why Cumulative CPI Stabilises (and Why That's Both Good and Terrifying)
Cumulative CPI stabilises because of simple arithmetic. As the denominator grows (more total EV, more total AC), each new month's data has a diminishing effect on the ratio. By month 6 on a 24-month project, the cumulative figures are large enough that even a terrible month barely moves the needle.
This is good because it gives you a reliable forecasting base. It's terrifying because it means recovery is mathematically difficult once cumulative CPI locks in. If your cumulative CPI is 0.91 at the halfway point, you'd need a period CPI of roughly 1.10 for every remaining month to pull the cumulative back to 1.0 by project end. On a construction project where 0.91 represents systematic inefficiency, not a one-off event, that kind of sustained improvement almost never materialises.
I tracked this on a £28M highways package in 2023-2024. Cumulative CPI hit 0.88 at month 8 of a 20-month programme. The team produced an ambitious recovery plan targeting CPI 1.0 for the remaining 12 months. Actual period CPI for the remaining months averaged 0.94. Final cumulative CPI: 0.91. Better than 0.88, but nowhere near the 1.0 they planned. The project finished £2.5M over target.
The CPI Stability Rule
Research from the US Department of Defense (Christensen, 1993, updated by multiple studies since) established that cumulative CPI stabilises within a narrow band of +/-0.10 from its final value once a project passes 20% completion. By 40% completion, the band tightens to +/-0.05.
For UK construction:
| Completion % | Cumulative CPI Reliability | Forecasting Implication |
|---|---|---|
| 0-15% | Low, still volatile | Don't use for forecasting. Too early. |
| 15-25% | Moderate, trend emerging | Start watching. Flag if below 0.92. |
| 25-40% | High, locked in +/-0.10 | Your final CPI will be within 10% of current cumulative CPI. Plan accordingly. |
| 40-60% | Very high, locked in +/-0.05 | Your final CPI is effectively known. If it's 0.88, your EAC should assume 0.83-0.93 range. |
| > 60% | Near-certain | Cumulative CPI at this point IS your final CPI. Recovery to 1.0 is a fantasy. |
Worked Example: Cumulative CPI Stabilising at 0.91
Worked ExampleScenario: A £22M NEC4 Option C water infrastructure project in the North East. The commercial manager has been tracking both cumulative and period CPI since month 1. At month 10 of a 20-month programme (50% duration elapsed, 46% physically complete), the data looks like this:
| Month | Period EV | Period AC | Period CPI | Cumulative EV | Cumulative AC | Cumulative CPI |
|---|---|---|---|---|---|---|
| 1 | £340K | £290K | 1.17 | £340K | £290K | 1.17 |
| 2 | £580K | £640K | 0.91 | £920K | £930K | 0.99 |
| 3 | £720K | £810K | 0.89 | £1,640K | £1,740K | 0.94 |
| 4 | £890K | £950K | 0.94 | £2,530K | £2,690K | 0.94 |
| 5 | £1,100K | £1,250K | 0.88 | £3,630K | £3,940K | 0.92 |
| 6 | £1,050K | £1,120K | 0.94 | £4,680K | £5,060K | 0.92 |
| 7 | £980K | £1,080K | 0.91 | £5,660K | £6,140K | 0.92 |
| 8 | £1,200K | £1,380K | 0.87 | £6,860K | £7,520K | 0.91 |
| 9 | £1,150K | £1,220K | 0.94 | £8,010K | £8,740K | 0.92 |
| 10 | £1,110K | £1,330K | 0.83 | £9,120K | £10,070K | 0.91 |
Period CPI range: 0.83 to 1.17. Bouncing wildly.
Cumulative CPI: Started at 1.17, dropped quickly, and has been oscillating between 0.91 and 0.94 since month 4. At month 10 (46% complete), it's 0.91.
What this tells the commercial manager:
Based on the stability rule, cumulative CPI is now locked in at roughly 0.91 (+/- 0.05). The realistic range for final CPI is 0.86 to 0.96.
EAC calculation:
- EAC = BAC / Cumulative CPI = £22M / 0.91 = £24.18M
- Forecast overrun: £2.18M against the target total of the Prices
On NEC4 Option C, that £2.18M overrun is shared through the pain/gain mechanism. If the share percentages are 50/50 above the target, the Contractor's share of the pain is £1.09M. That's real margin erosion.
The commercial manager's next step: investigate the root causes of the persistent inefficiency. Month 5 and month 10 both showed period CPI below 0.88, what drove those? If it's compensation events that haven't been notified, the BAC should increase, which would improve cumulative CPI retrospectively.
The Forecasting Power of Cumulative CPI
Every common EAC formula uses cumulative CPI as an input:
| EAC Formula | When to Use | Cumulative CPI Role |
|---|---|---|
| EAC = BAC / CPI | Past efficiency will continue | CPI is the sole driver |
| EAC = AC + (BAC - EV) / CPI | Same as above, explicitly showing remaining work | CPI adjusts remaining cost |
| EAC = AC + (BAC - EV) / (CPI x SPI) | Both cost and schedule inefficiency will continue | CPI combines with SPI |
| EAC = AC + Bottom-up ETC | Manager's estimate for remaining work | CPI used as a sanity check against the bottom-up figure |
The bottom-up ETC (manager's estimate) deserves special mention. On UK construction projects, most commercial managers forecast final cost using a bottom-up package-by-package estimate. Not a formula. That's fine. But check it against the CPI-based EAC. If your bottom-up forecast says £23.5M but the CPI-based EAC says £24.2M, someone is being optimistic. Usually it's the bottom-up estimate. The £700K gap is a conversation you need to have at the commercial review, not discover at final account.
Common Mistakes
- Using period CPI for EAC calculations. Period CPI is volatile. Plugging this month's period CPI of 0.78 into an EAC formula produces a catastrophic forecast that probably won't materialise. Use cumulative CPI for trend-based forecasting. Period CPI is useful for diagnosing what happened this month. Not for predicting the future.
- Assuming cumulative CPI will recover to 1.0. After 25% completion, cumulative CPI recovers to 1.0 only if the project achieves sustained period CPI well above 1.0 for every remaining month. On construction projects with embedded inefficiencies (poor productivity, subcontractor issues, design problems), that's fantasy. Plan around the current cumulative CPI, not the one you wish you had.
- Not separating CE impacts from organic inefficiency. A cumulative CPI of 0.91 might be 0.96 after adjusting the baseline for compensation events that should have been implemented. If your BAC is wrong because CEs haven't been priced and added, your cumulative CPI is artificially low. Fix the baseline first, then assess the real efficiency.
- Ignoring the first 3 months of data. Some teams don't calculate CPI until month 4 because "the data is too volatile." That's understandable but wasteful. Record it from month 1. You won't trust it until month 4, but you'll want the historical trend line when you're analysing performance at month 12.
How Gather helps. Gather's AI reads your site diaries daily and maps progress against your cost-loaded programme, giving you accurate earned value data without manual spreadsheet updates. Book a demo to see it working on a live NEC4 project.
Frequently Asked Questions
What is cumulative CPI in earned value?
Cumulative CPI is the Cost Performance Index calculated from project start to the current reporting date: total EV divided by total AC across all periods. It's more stable than period CPI because it averages out monthly fluctuations. Cumulative CPI is the primary input for trend-based EAC forecasting, once it stabilises (typically around 20-25% project completion), it predicts final cost performance with surprising accuracy.
Why is cumulative CPI more reliable than period CPI?
Because it averages out the noise. Period CPI reacts to every large invoice, every subcontractor payment spike, and every month where progress was measured slightly differently. A single large material delivery can tank period CPI one month and then it bounces back the next. Cumulative CPI absorbs these swings and reveals the underlying efficiency trend. After 4-6 months of data, the cumulative figure is far more trustworthy for forecasting.
Can cumulative CPI recover from a poor start?
Mathematically, yes. Practically, rarely. Research across defence and infrastructure projects shows that cumulative CPI stabilises within +/-10% of its final value by 20% project completion. If cumulative CPI is 0.88 at 25% complete, recovering to 1.0 would require sustained period CPI of roughly 1.04 for every remaining month. On projects where 0.88 reflects genuine inefficiency rather than a one-off event, that improvement almost never materialises. The realistic approach: forecast around 0.88, identify compensation events that should adjust the baseline upward, and focus recovery efforts on preventing further deterioration.
How do you calculate cumulative CPI from period data?
Sum all period EV values to get cumulative EV. Sum all period AC values to get cumulative AC. Divide cumulative EV by cumulative AC. Don't average the period CPI values. That gives a different (incorrect) result because the periods have different magnitudes. For example, if month 1 has EV/AC of £100K/£90K and month 2 has £500K/£550K, the cumulative CPI is (£100K + £500K) / (£90K + £550K) = £600K / £640K = 0.938. Not the average of 1.11 and 0.91, which would be 1.01.
.webp)
