Measured mile and disruption
Last reviewed 16 September 202612 min read
Why it isn't delay, how the comparison works, and the six records that decide whether you can run it at all
🟩 In one line Disruption is loss of productivity — more effort for the same output — whether or not the project finishes late. The measured mile quantifies it by comparing an unimpacted period of the same work, same crew, same project against an impacted one. Its strength is that both halves are actual. Its weakness is that it needs a clean period to exist.
👤 Who this is for Planners on congested, heavily varied or accelerated jobs, who need to know what to start recording now. Commercial and claims staff choosing a quantification method. Senior planners being asked why the programme shows on-time completion while the labour account is 40% over.
First, let's be honest about why this page exists
Because disruption is the loss that gets discovered too late to prove.
Delay has a natural home. It lives in the programme, the programme is updated monthly, and by the time anybody argues about it there are thirty archived files describing what happened. Disruption has no such home. Nothing in the schedule records that a crew of twenty took forty percent more hours than it should have to finish on the planned date. The bars are the same length. The milestones are green. The money is gone.
And unlike delay, disruption cannot be reconstructed afterwards. You can re-run a programme. You cannot re-run a foreman's headcount in April.
So the honest framing of this page is: most of the value is in the last section, not the first. The arithmetic of a measured mile takes four lines. Whether you will ever be able to run one is decided in the first month of the project, by whether somebody set up labour allocation against the same codes the programme uses. If you read only one part of this page, read the record set.
This page covers the measurement technique only. Whether you are entitled to recover disruption costs is a contractual and legal question, and it is not addressed here.
🟨 The standard — what "good" looks like
Disruption is a loss of productivity — the same output achieved with more resource, or less output achieved with the same resource — caused by an event or condition for which the claiming party is not responsible. It is distinct from delay, which is loss of time.
The measured mile quantifies disruption by comparing productivity in an unimpacted ("clean") period with productivity in an impacted period, using actual records from the same project and, ideally, the same crew and work type.
The recognised position across the main sources:
| Source | What it establishes |
|---|---|
| AACE RP 25R-03, Estimating Lost Labor Productivity in Construction Claims | Sets out the hierarchy of quantification methods and places project-specific studies — the measured mile at the top — above comparable-project studies, above industry factor tables, above total-cost approaches. |
| SCL Delay and Disruption Protocol, 2nd ed. | Treats disruption as separate from delay, and expresses a clear preference for project-specific measured approaches over estimated or modelled ones. |
| AACE RP 29R-03, Forensic Schedule Analysis | Establishes what delay analysis does and does not answer — which is why a delay analysis, however good, will not quantify disruption. |
| GAO-16-89G, Schedule Assessment Guide | Activity-level detail and record standards, which determine in practice whether productivity can be compared at all. |
🟢 Rule: pick a representative unimpacted period, not the best one; hold work type, crew, location and access constant across both halves; make both halves long enough to average out noise; and be able to show, from records made at the time, that the clean period was actually clean. If any one of those four cannot be evidenced, the measured mile is not available and you drop down the hierarchy.
How it actually works
Delay and disruption are different questions
Both can arise from the same event. Neither proves the other.
| Delay | Disruption | |
|---|---|---|
| The question | When did it finish, and why late? | How much more effort did it take? |
| Unit | Days | Man-hours, or output per man-hour |
| Evidence | Programme updates, driving path | Labour allocation and quantity records |
| Requires late completion | Yes, generally | No |
| Primary method | Schedule analysis | Productivity comparison |
The calculation
Three lines, and every input is an actual:
- Unimpacted productivity = quantity installed ÷ man-hours expended, in the clean period
- Impacted productivity = quantity installed ÷ man-hours expended, in the affected period
- Lost man-hours = man-hours actually expended in the impacted period − (impacted quantity ÷ unimpacted productivity)
Note what is not in there: the tender rate, the planned productivity, an industry study, or anybody's opinion about what was reasonable. That absence is the entire point. The comparison removes almost every argument about whether the planned rate was ever achievable, because it never uses the planned rate. It uses what this crew, on this project, actually achieved when conditions were normal.
What makes a period genuinely clean
Five conditions. All five have to hold, and the other side will test each one individually.
- Same work type. Blockwork against blockwork. Not blockwork against general finishing.
- Same or comparable crew. A different subcontractor, or the same one after significant turnover, is a different comparison.
- Same or comparable location and access. Ground floor against floor 34 is not fair — height, hoisting and congestion all differ, and the difference is not the disruption you are claiming.
- No significant impacts in the period. No major variation, no acceleration, no severe congestion, no abnormal weather.
- Long enough to be representative. A single week captures noise. Several weeks, or a complete repeating cycle — one full floor on a typical tower — captures performance.
When there is no clean period
It happens more often than the textbooks suggest, because on badly disrupted jobs the disruption starts in month two and never stops. The measured mile is then unavailable and you drop down the hierarchy:
| Rank | Method | What it uses | Weakness |
|---|---|---|---|
| 1 | Measured mile | This project, this crew, both periods actual | Needs a clean period |
| 2 | Comparable work or project | Similar scope elsewhere, with records | Comparability is challengeable |
| 3 | Industry studies and factor tables | Published research (MCAA, NECA, Business Roundtable) | Conditions may not match yours |
| 4 | Earned value comparison | Planned against earned hours | Depends on the planned figure being reasonable |
| 5 | Total cost / modified total cost | Overrun as the measure of loss | Assumes the estimate was right and you were faultless |
The bottom of that list is treated with considerable scepticism, and reasonably so: it asks the reader to accept that the tender was correct and that none of the overrun was the claimant's own doing. Modified total cost improves on it by deducting the claimant's own contribution, but the burden remains uncomfortable.
Everything above rank 3 requires project records at activity level — which is where this page ends up, and why.
🟥 Where people go wrong
1. Picking the best period instead of the normal one. This is the mistake that discredits otherwise sound analyses. If the selected clean period is the crew's single best fortnight of the project, the comparison measures a peak against an average and the loss is overstated. Choose a representative period and be able to explain in one sentence why it is representative.
2. Booking labour against activities too coarse to compare. Hours booked against a 90-day activity spanning six floors and three trades cannot be compared with anything. Booked against Blockwork L07 Zone 2, they can. This is why activity duration limits is a commercial decision taken at baseline, not a scheduling preference.
3. Running a delay analysis and calling it a disruption case. They answer different questions from different records. A flawless time impact analysis establishes entitlement to time. It establishes nothing at all about lost man-hours.
4. Quantities and hours on different breakdowns. The quantity surveyor measures by bill item; the foreman books hours by area; nobody can divide one by the other. If the two are not on the same code structure, the measured mile is arithmetically impossible no matter how complete both records are.
5. Starting the records when the dispute starts. By definition the clean period is early and the impacted period is later. If allocation sheets begin in month fourteen because that is when somebody got worried, the clean half of the comparison does not exist and never will.
6. Claiming the whole productivity gap. Some of the loss will be the claimant's own — learning curve, poor supervision, absenteeism, rework. An analysis that attributes 100% of the gap to the other side invites the tribunal to discount the whole thing. Deduct what is yours and say that you have.
7. Treating the percentage as the claim. "35% productivity loss" is a finding, not a number to invoice. The claim is lost man-hours times a rate, with the rate evidenced, and with the causal link to specific events established separately.
📜 What the measurement does and does not decide
A measured mile quantifies a loss. It does not establish that anyone owes you for it. Three things sit between the number and the money, and all three are contractual rather than technical.
- Cause. The analysis shows productivity fell. It does not show why. The causal link runs through the variation log, the RFI register, the instruction record and the site diaries — and it has to be built event by event, not asserted globally.
- Entitlement. Whether the cause is one the contract makes recoverable. Under FIDIC this typically runs through the variation and claims machinery rather than the time clauses; under NEC it runs through compensation events, where the assessment is forward-looking and the mechanism is materially different.
- Notice. The same notice provisions that govern time claims usually govern cost claims too, and they are frequently condition-precedent. A well-evidenced disruption loss that was never noticed can be worth nothing. See notice requirements and records.
The practical consequence for a planner: the moment you can see productivity degrading in the labour returns — not when the claim is being assembled — is the moment to tell the commercial team, because that is when the notice clock is running.
General guidance, not legal advice. Contract wording, editions and amendments differ on every project. Read your own contract and take professional advice before relying on anything here in a claim or a dispute.
⚖️ When you're challenged
"Your clean period was the best month on the job." "It was the third of five typical-floor cycles, and it sits within 4% of the mean of the other four unimpacted cycles. I've shown all five. If you'd prefer I use the mean of all of them rather than that single cycle, the loss reduces by about 2% and the conclusion doesn't change." Having that alternative ready in advance is what closes this line of attack.
"The project finished on time, so there's no claim." "Completion is a delay question. This is a productivity question. The building finished on the contract date because we ran 1,180 more man-hours than the sequence required, and those hours are the loss."
"Some of that is your own inefficiency." "Agreed, and I've deducted it. The learning-curve allowance on the first cycle and the two weeks of subcontractor undermanning in March are excluded from the impacted period. What's left is the 1,186 hours attributable to the access restrictions." Concede the concedable point first; it makes the rest credible.
"Why not just use the MCAA factors? Everyone does." "Because the factors are a fallback for when project records don't exist. We have the records. A study of this crew on this building is stronger evidence than a published average for a different market, and it's harder to argue with."
"Where does the man-hour figure come from?" "Daily allocation sheets, signed by the foreman, coded to the same WBS the programme uses. Here's the sheet for 14 April and here's the same day in the quantity return. Every number in the calculation traces back to one of those two documents." If you cannot say that sentence, you do not yet have a measured mile.
📄 Related pages
- Manhour budget and productivity tracking — the live system that produces the records this method needs
- Notice requirements and records — why prompt notice decides whether the number is worth anything
- Activity duration limits — why activity size determines whether productivity can be compared at all
- Progress measurement and rules of credit — getting quantities onto the same basis as the hours
- Choosing a delay method — the time side, and why it won't do this job
- Productivity adjustment factors — the rank-3 fallback when no clean period exists
- Productivity rates — published and field rates, for sanity-checking either half
✏️ Worked example
Blockwork package, Abu Dhabi residential tower. Typical floors L05–L24, one floor per cycle. Records: daily labour allocation coded to BLK-L‹nn›-Z‹n›, and weekly quantity returns on the same codes.
Clean period. L07–L10, four consecutive cycles, no variations affecting the floors, single subcontractor, no hoist restrictions.
| Cycle | Quantity (m²) | Man-hours | m² / mh |
|---|---|---|---|
| L07 | 1,040 | 760 | 1.37 |
| L08 | 1,060 | 745 | 1.42 |
| L09 | 1,050 | 760 | 1.38 |
| L10 | 1,050 | 735 | 1.43 |
| Total | 4,200 | 3,000 | 1.40 |
Impacted period. L16–L19. Façade installation running concurrently on the same floors following a design change; single hoist shared with three trades; blockwork gangs working in half-floor sections with daily relocation.
| Cycle | Quantity (m²) | Man-hours | m² / mh |
|---|---|---|---|
| L16 | 790 | 850 | 0.93 |
| L17 | 760 | 860 | 0.88 |
| L18 | 780 | 840 | 0.93 |
| L19 | 770 | 850 | 0.91 |
| Total | 3,100 | 3,400 | 0.91 |
The calculation:
- Man-hours that should have been needed = 3,100 ÷ 1.40 = 2,214
- Man-hours actually expended = 3,400
- Lost man-hours = 3,400 − 2,214 = 1,186
- Productivity loss = 1,186 ÷ 3,400 = 34.9%
Now the part most write-ups skip. Of those 1,186 hours, the labour returns show that in cycle L17 the subcontractor was two masons short for six days for reasons unconnected to the façade works. That accounts for roughly 96 hours. Deduct them.
- Attributable lost man-hours = 1,090
Deducting your own 96 hours costs you 8% of the claim and buys you the credibility of the other 92%. It is almost always the right trade.
The paragraph that goes in the submission:
Blockwork productivity in the unimpacted cycles L07–L10 averaged 1.40 m² per man-hour across 4,200 m² and 3,000 man-hours. In cycles L16–L19, during concurrent façade installation on the same floors, productivity fell to 0.91 m² per man-hour across 3,100 m² and 3,400 man-hours. Applying the unimpacted rate to the impacted quantity gives 2,214 man-hours, against 3,400 expended — a shortfall of 1,186 man-hours. 96 man-hours attributable to Contractor-side undermanning in cycle L17 are deducted, giving 1,090 man-hours. All figures are taken from daily labour allocation sheets and weekly quantity returns coded to the same WBS, copies of which are at Appendix C.
Six sentences. Every figure traceable to a document made at the time. That last sentence is what the whole method rests on.
📖 References
- AACE International RP 25R-03, Estimating Lost Labor Productivity in Construction Claims — the method hierarchy, the measured mile, and the conditions for a valid comparison period
- SCL, Delay and Disruption Protocol, 2nd ed. (February 2017) — disruption as distinct from delay, and the preference for project-specific measured approaches
- AACE International RP 29R-03, Forensic Schedule Analysis — why delay analysis and disruption analysis answer different questions
- US GAO, Schedule Assessment Guide (GAO-16-89G, 2015) — activity-level record standards that determine whether productivity comparison is possible
- Mechanical Contractors Association of America, Change Orders, Productivity, Overtime — the sixteen factors, as the rank-3 fallback
From the field
Notes, corrections and additions from people who've done this on a real job. Everything is checked before it appears here.
Loading notes…
Checked before it appears. No client, project or employer names. Your name is shown; your email never is. Terms.