Earned Value Common Mistakes in Construction: 12 Pitfalls That Wreck Your EVM Data

Earned value management should be the sharpest tool in a commercial team's kit. In practice? Most construction teams get it wrong. Not slightly wrong. Wrong enough that the CPI they're reporting to the board bears almost no resemblance to what's happening on site.

I've watched EVM rolled out on projects worth £200M+ where, six months in, the commercial team couldn't explain why their CPI said 1.05 but the project was haemorrhaging cash. The formulas weren't wrong. The inputs were.

Here are twelve earned value common mistakes in construction — what they look like, why they happen, and how to fix them before they wreck your reporting.

1. Measuring Progress by Percentage Guess Instead of Physical Measurement

This is the single most common EVM mistake in construction. It's also the most damaging.

Every month, the site team reports progress as a percentage. "Foundations are 60% complete." But 60% of what? Nobody's actually measured the cubic metres poured against the total required. They've eyeballed it. Maybe the foreman said "about 60%." Either way, the earned value figure is fiction dressed up as data.

Percentage estimates are reliably optimistic. There's decades of research on this — look up the 90% syndrome, where tasks sit at "90% complete" for months. It happens because physical measurement is slow, unglamorous work. Percentages are faster. But they're not EVM.

How to fix it

Define physical measurement rules for every cost-significant WBS element before the project starts. Not guidelines. Rules.

If you can't physically measure it, use milestones with weighted values. "Foundations" becomes: excavation complete (20%), blinding poured (5%), rebar fixed (30%), concrete poured (35%), curing and backfill (10%). That's still imperfect, but it's miles better than a foreman's gut feeling.

2. Baselining Against an Unrealistic Programme

Your earned value baseline is only as good as the programme it sits on. If the programme was written to win the tender rather than to deliver the project, your planned value curve is fantasy from day one.

I've seen this repeatedly on NEC4 Option C packages. The Accepted Programme gets agreed with ambitious logic, the PV curve tracks that ambition, and within three months the SPI is screaming red — not because the team is underperforming, but because the baseline was never achievable.

The worst version: refusing to re-baseline after compensation events that change the completion date. Under NEC4, a compensation event doesn't just change the Prices — it changes the Accepted Programme. If you don't update your PV baseline to match, your schedule variance is measuring against a target that no longer exists.

How to fix it

• Validate the baseline against resource availability, not just logic
• Under NEC4, re-baseline after every implemented compensation event that changes planned Completion
• Keep the original baseline as "Baseline 0" for historical comparison, but measure current SV/SPI against the latest approved baseline
• Test it: if a CPI of 1.0 would require productivity rates 30% above what the supply chain has ever achieved, the baseline is wrong

3. Confusing Committed Cost with Actual Cost

Your earned value actual cost (AC) should reflect the cost of work actually performed in the period. Not what you've committed in subcontract orders. Not what you've been invoiced. What you've spent on work done.

A subcontractor has a £2M order for structural steelwork. In month 4, they've erected £400,000 worth of steel on site. But the QS has booked £800,000 as actual cost because that's what the subcontractor invoiced — front-loaded, as subcontractors always are. The CPI now shows 0.50, a project in apparent crisis, when the real performance is 1.0.

The reverse happens too. Teams book actual cost only when an invoice arrives. Labour working in January gets invoiced in March. January's ACWP is understated, CPI looks excellent, everyone relaxes. Then March hits and the numbers collapse.

How to fix it

• Accrue labour weekly based on timesheets, not payroll dates
• Value subcontractor work monthly based on physical progress, not invoices
• Book materials when installed, not when delivered or paid for
• Keep a separate reconciliation between EVM actual cost and financial accounts

This is harder than it sounds. But if your AC doesn't match the timing of your EV, every metric you calculate is meaningless.

4. Ignoring Scope Changes That Move the Target

On an NEC4 Option C target cost contract, every compensation event potentially changes your Budget at Completion (BAC). If you don't update BAC when the target changes, your CPI is measuring performance against the wrong number.

A £45M highways project starts with a BAC of £45M. Six months in, 15 compensation events have added £3.2M to the target. But nobody's updated the BAC. Because EV = % complete × BAC, an outdated BAC produces a wrong EV — and a wrong CPI. The team reports CPI 0.92 and panics. Recalculate against the correct BAC of £48.2M and CPI is actually 0.98. The project isn't failing. The measurement was.

How to fix it

Tie BAC updates to the CE implementation register. Every compensation event implemented under clause 65.1 should trigger a BAC revision.

Update BAC monthly as part of the period close, not quarterly. Quarterly is too slow — you'll be reporting three months of wrong data.

5. Treating All Variance as Recoverable

A CPI of 0.90 in month 3 doesn't mean the same thing as a CPI of 0.90 in month 18.

Early in a project, cost overruns might genuinely be recoverable. But late in the project, variance is almost always baked in. The worst version is the "hockey stick" forecast — a current CPI of 0.85 but an EAC that assumes the remaining work gets done at CPI 1.10. I've lost count of how many times I've seen this in project board packs. That's not a forecast. That's hope dressed up in a spreadsheet.

How to fix it

Use different EAC formulas for different project stages:

Never accept an EAC where the implied CPI for remaining work is more than 10% above the cumulative CPI to date. "We'll try harder" isn't a plan.

6. Not Updating the EAC Frequently Enough

Some teams calculate EAC quarterly. On a 24-month project, that's eight data points. You wouldn't drive a car checking the speedometer eight times during a journey.

A project drifting 2% per month doesn't look alarming in any single period. Nobody panics over 2%. But skip EAC for six months and you've got a 12% overrun that nobody saw coming. On a £30M package, that's £3.6M. By the time quarterly EAC catches it, the money's gone.

How to fix it

Build EAC into the monthly commercial cycle. Period close isn't finished until EAC is updated. Don't just run the formula-driven EAC — include the bottom-up ETC from the commercial team's assessment of remaining work. Compare the two. If they diverge by more than 5%, investigate before reporting. That gap between the formula and the professional estimate is where the truth usually hides.

7. Cherry-Picking Metrics: Reporting CPI but Hiding SPI

A project reports CPI of 1.02 to the board. Looks great. On budget. What they don't mention is SPI of 0.78. The project is 22% behind programme. They're on budget because they haven't done the work yet. When they do catch up on the schedule variance, the CPI will follow the SPI down — and it won't be pretty.

The reverse happens too. A project reporting excellent schedule performance (SPI 1.05) as evidence of good management while CPI is 0.88 — because they're ahead of programme precisely because they're throwing money at acceleration. The two metrics are connected. Reporting one without the other is misleading.

How to fix it

Always report CPI and SPI together. Use the Cost Schedule Index (CSI = CPI × SPI) as a combined health metric. A CSI below 0.90 should trigger a formal recovery plan regardless of what the individual metrics show.

Make it a standing question in every project review: "What's your CPI, SPI, and CSI this period?" Silence on SPI is usually louder than the CPI figure they did share.

8. Poor WBS Structure That Doesn't Match How Work Is Managed

The classic mistake: building the WBS around the bill of quantities rather than around how work is actually managed on site. A BoQ might have 400 line items for a concrete frame. The site team manages pours — ground floor slab, first floor columns, first floor slab. If your WBS tracks BoQ lines, you'll spend more time allocating costs than managing them.

Then there's the opposite problem: WBS elements too large to be useful. A single £5M control account for "Mechanical Installation" tells you nothing. When the CPI drops, is it pipework, ductwork, or the plant room fit-out? You can't act on what you can't see.

How to fix it

Target control accounts of £250,000–£500,000 for a typical £30–50M project. Each one needs a single responsible person, a defined scope of physical work, a measurable output (see Mistake 1), and a budget from the estimate — not from the BoQ.

Agree the WBS before setting the baseline. Changing it mid-project is painful — you lose all your trend data and have to re-map months of cost allocation. Get it right early, as part of your EVM implementation.

9. Not Integrating EVM with the NEC4 Programme

This is the mistake specific to UK construction. Under NEC4, the Accepted Programme (clause 31) is the contractual baseline for time and, under Options C/D/E, for forecast cost. If your EVM planned value curve doesn't align with the Accepted Programme, you have two competing versions of what "on plan" means.

I've seen this on a £60M water treatment project where the EVM baseline was set at tender and never reconciled with the Accepted Programme revisions. By month 12, the two baselines had diverged by four months. The PM thought the project was two months late. The EVM said it was two months early. Both were "correct" against their own baselines. Neither was useful.

How to fix it

The PV curve must be derived from the current Accepted Programme. When the programme revises under clause 32, the PV curve gets revised too. In practice: export the resource-loaded Accepted Programme as a cost-phased spend profile, use that as the PV curve, and regenerate it with each accepted revision. Visualise alignment with an S-curve — plotting PV, EV, and AC cumulatively makes any divergence between the curves immediately obvious.

10. Applying Manufacturing EVM Rules to Construction Without Adaptation

EVM was developed for US defence manufacturing in the 1960s: stable scope, controlled factory conditions, predictable productivity. Construction is none of those things. Three manufacturing assumptions that don't survive contact with a building site:

• Level of Effort as a fixed percentage of direct work — Factory overheads are stable. Construction preliminaries swing wildly when the programme extends or the client demands acceleration
• One baseline for the whole job — Manufacturing scope is locked. Construction scope shifts constantly through compensation events, variations, and design development
• Monthly reporting is enough — A factory runs continuously. A construction site has weather losses, August shutdowns, and peak/trough trade cycles

How to fix it

Analyse CPI and SPI by phase, not just cumulatively. Track preliminaries as a separate cost stream — they behave completely differently to measured work. Re-baseline at major scope changes (more than 10% change warrants a new comparator). Weight phases by risk, not just by cost. The earned value construction example walks through these adaptations in detail.

11. Over-Reliance on Formulas Without Professional Judgement

I watched a project team report a CPI of 1.08 for eight consecutive months. The commercial director was delighted. What the formula didn't capture: the team was deferring major plant demobilisation costs, booking subcontractor retention as a cost saving, and hadn't accrued for known defects. The real CPI was closer to 0.92. The "data-driven" reporting was hiding a £1.6M problem.

Formulas can't tell you about costs that haven't been booked yet but are certain to arrive, quality issues that will require rework, subcontractor claims being negotiated, or acceleration costs buried in future periods.

How to fix it

Every monthly EVM report should include a "QS overlay" — a paragraph from the commercial manager stating what the formulas don't capture. Risks, known but unbooked costs, commercial disputes, pending claims. The narrative is where the real intelligence lives. Encourage your team to challenge the numbers: if the CPI says everything's fine but the site team is working weekends, something's wrong with the data, not the site team.

12. Failing to Act on Early Warning Signals

The entire point of earned value management is early warning. And yet, project after project, the team watches the CPI drift from 1.0 to 0.98 to 0.95 to 0.91 over four months and does nothing. They wait for it to "come back."

It doesn't come back.

Under NEC4 clause 15, both the Project Manager and Contractor must notify early warnings about anything that could increase the total of the Prices. EVM gives you exactly that signal. A CPI trending below 1.0 over three consecutive periods isn't just data — it's telling you something is wrong. Treating these signals as dashboard decoration defeats the entire purpose of running EVM in the first place.

How to fix it

Set explicit trigger thresholds with named response owners:

Don't just set the thresholds. Assign names to the responses. "If CSI drops below 0.90, [name] convenes a recovery workshop within 48 hours." Otherwise the triggers are just decoration on a dashboard nobody reads.

Worked Example: How Three Mistakes Compound on a Real Project

Quick Reference: 12 Earned Value Mistakes at a Glance