Construct IQ

The Cost of Rework in Construction: What It Costs, Where It Starts and How to Measure It

By Kush Agarwal, President and CEO, YelloSKYE. Published 2026-10-29.

Executive Summary

• The Question: How much does rework cost a construction project, where does it come from, and how can a team put a reliable number on it?
• The Answer: The direct correction usually takes a few per cent of project value; once delay, supervision, remobilisation and disputes are added, the total is often several times larger. Most errors start with information and trade interfaces rather than workmanship, and the price of each one depends mainly on how long it stays hidden.
• The YelloSKYE Angle: A dated, located record of the site, captured through Construct IQ and held on DEX, shows when a faulty element went in and when it was found, which is the input most rework measurement is missing.

Rework is effort spent removing, correcting or repeating work that did not meet the drawings, specification or agreed intent the first time it was built. Research puts its direct cost at around 5 per cent of construction cost on average, and broader estimates that count the knock-on effects reach 10 to 25 per cent. The true figure on any one project is usually unknown, because few teams record rework as a cost of its own.

This guide deals with the size of the problem: what counts as rework, which costs follow a single error, where errors begin and how to measure them on your own sites. Prevention has its own companion guide on how to reduce rework. Both draw on how YelloSKYE handles site evidence in Construct IQ, the construction monitoring part of our platform.

What counts as rework

A workable definition: any work that has to be undone or done again because it was not right when installed, where the correction is not covered by an approved change to the scope. Three neighbouring categories are often lumped in with it, and keeping them apart matters because each is paid for differently.

The line between rework and change is where most of the money is contested. If a wall comes down because the tenant layout moved, that is a change. If it comes down because it was set out from a superseded drawing, it is rework, and the argument turns to who issued, received and used the old revision. Snagging is covered in our construction punch list guide; this one stays with errors that force finished work to be undone while the project is still being built.

What the research says about the size of the problem

Published estimates differ because they count different things, but they point the same way.

• Construction Industry Institute (CII). Its field studies, mostly of industrial projects in the US, found that direct field rework averaged about 5 per cent of construction cost. That covers only the visible correction on site.
• Get It Right Initiative (GIRI). This UK industry research group took a wider view, adding the indirect effects of avoidable errors, and put their cost at somewhere from 10 to 25 per cent of a project's value.
• FMI and PlanGrid. Their 2018 study, Construction Disconnected, estimated that rework caused by poor project data and miscommunication cost the US industry $31.3 billion in a single year. The same survey found that construction professionals spent more than 14 hours a week on non-productive work such as searching for information, resolving conflicts and dealing with mistakes.

Taken together, the studies say one thing in different units: the correction you can see on site is the smaller part of the bill.

The full cost of a single rework event

Take an ordinary error. A row of floor sleeves is cast in the wrong position, and nobody notices until the plumbing contractor arrives with prefabricated risers. These are the costs that follow.

Only the first two rows are normally recognised as rework, and even those are often booked to ordinary labour and material lines with no cause attached. The rest is real money that surfaces elsewhere in the accounts, which is why rework measured from cost reports alone nearly always looks smaller than it is.

Who ends up paying

The party that pays is often not the party that made the mistake. Under a fixed-price contract, the main contractor usually absorbs the correction first and then tries to recover it from the responsible trade or designer. On cost-reimbursable work, much of it can pass through to the owner unless the contract says otherwise. Subcontractors on lump sums carry their own corrections and frequently the disruption caused by other trades' errors as well. Owners pay less visibly, through late completion, delayed revenue and management time spent on claims.

Moving any of that cost to the party responsible depends on evidence of what was built, when and to which drawing. Our construction dispute documentation guide explains what that evidence needs to look like.

Where rework comes from

Rework tends to be blamed on whoever did the work. Root-cause studies point elsewhere: most errors are set up before anyone picks up a tool.

The first four rows are information failures. The crew built correctly to what it was given, and what it was given was wrong, out of date or incomplete. The FMI and PlanGrid research attributed close to half of all rework to problems of this kind. The practical lesson is that better tradespeople alone will not move the number much; getting the right information to the right crew will.

Why the price depends on when the error is found

The biggest single driver of what a rework event costs is detection lag: the time between an error being built and someone noticing it. During that interval, other work goes on top of it.

Consider a drainage run laid at the wrong fall below a ground-floor slab. While the trench is open, re-bedding the pipe takes an hour or two. Once the slab is cast over it, the fix means breaking out concrete, relaying the pipe and recasting, with the flooring contractor waiting. If the problem only appears when the drain backs up after occupation, the owner's tenants are disrupted too. The error never changed; the date of discovery set the bill.

YelloSKYE's as-built versus BIM clash detection work shows the same pattern on a structure. A column or shaft slightly out of position looks harmless until the MEP contractor's prefabricated runs fail to fit weeks later. Laser scanning after each critical structural stage, overlaid on the design model, moves that discovery back to the point where the fix is still cheap.

How to measure rework on your own projects

Industry averages are a reference point, not a target. The number that matters is your own, and most organisations lack it because rework has no natural place in a standard cost structure. Measuring it takes four things.

1. One definition everybody uses

Write down what counts, using the categories above, and make sure site teams, commercial staff and trades apply the same test. Otherwise one site logs every snag as rework and the next logs nothing at all.

2. A log entry for every event

Record each event when it is found, not at month end from memory. The minimum fields are:

• Location and element
• Date found and, as closely as possible, the date the faulty work was installed
• Source category, from the table above
• Labour hours and materials to remove and redo the work
• Days of delay to the area and to the trades that follow
• Responsible party, where it is known
• Evidence: images or scans before and after the correction

3. A cost code or cause tag

Give rework its own cost code, or add a cause tag to existing labour and material codes, so the money stops vanishing into general variance.

4. A short set of monthly metrics

Detection lag is the metric most teams cannot calculate. They know when an error was found, but not when it was built. Filling that gap needs a site record with dates on it.

Dating errors with a visual site record

When the site is captured on a fixed cadence, each rework event can be traced back through the record to the first capture in which the faulty work appears. That gives the installation date, the trades active in that area at the time and, often, enough context to establish which drawing was being followed. Detection lag becomes a measured number rather than a guess, and the commercial team has the evidence it needs to place the cost with the right party.

Construct IQ combines four sensors: drones, fixed cameras, 360° cameras and laser scanners. DEX keeps all captures in date order, letting reviewers step back through the site history day by day, lay CAD or BIM models over the site imagery to see where built work departs from the design, and tag an issue at its exact location so evidence and finding stay together. The same record does other jobs too. Flying the whole site once a week, as in the construction progress capture use case, returned more than 15 hours a week to site managers who had been compiling reports by hand. When resolving contractor billing disputes, drone flights timed to each billing cycle cut the site visits needed to verify claims by 80 per cent.

How often you capture sets how precisely you can date an error. Weekly drone flights place an external error within a week. A fixed camera on a busy zone narrows it to hours. A 360° walk just before ceilings close records each room at the last moment its services can be seen.

Why rework stays under-reported

• No cost code. Corrections are booked to the same lines as the original work.
• Incentives to say nothing. Reporting your own rework invites a backcharge or awkward questions.
• Indirect costs land elsewhere. Delay and supervision are rarely linked to the event that caused them.
• Small events go unlogged. Each seems too minor to record, but dozens add up.
• Records are pieced together late. By the final account, nobody can say which delay came from which error.

A 30-day plan to start measuring rework

1. Agree a one-paragraph definition of rework and the source categories, and brief site and commercial teams on both.
2. Add a rework cost code or cause tag to your cost system.
3. Set up a simple log with the fields listed above, stored where site, commercial and design teams can all see it.
4. Check that the site is captured often enough, for the pace of work, to date each event.
5. At the end of the month, calculate the rework cost ratio, events by source and average detection lag.
6. Take the largest source category into your prevention plan, using the steps in the companion guide, how to reduce rework.

The first month of honest figures is rarely comfortable reading. It is also the only baseline against which any prevention effort can be judged.

Explore Construct IQ