What counts as rework, and what is just drafting?
Rework is work that produces something which already existed in some usable form. Not iteration, which is thinking getting sharper. Rework is version two existing because version one was the wrong shape, pulled the wrong figures, used last year's wording, or sat in a format nobody could sign off.
The distinction matters because firms defend rework as craft. A partner marks up a report heavily and calls it quality control. Sometimes it is. But if the same marks appear on every report from every author, that is not quality control. That is a document being written twice because the standard was never written down once.
A simple test: could the second version have been produced first, if the person drafting had known what the reviewer knows? If yes, you are looking at rework. If no, because the client changed their mind or a new fact emerged, you are looking at real work.
Why does the same document get redone?
The causes repeat across firms of every kind:
- The wrong starting point. Someone finds a previous document that looks close enough, copies it, and spends an hour hunting for the parts that belong to the old job. Half the time they miss one.
- Unwritten house style. Each reviewer has preferences about structure, tone, length and what goes in an appendix. None of it is recorded, so every author guesses, and every guess gets corrected.
- Scattered source material. The figures sit in one system, the correspondence in another, the site or file notes in someone's inbox. Assembling them is slow, so people work from memory and then correct themselves later.
- Templates that have drifted. The master was sound two years ago. Since then it has forked into personal copies, each with a different clause set. We set out how that happens in why templates rot.
- Late review. A document reviewed three days after drafting gets rewritten. The same document reviewed the same afternoon gets corrected. The gap turns small fixes into large ones.
None of these is a people problem. They are all design problems in how the work flows.
Where does rework hide in an ordinary week?
It hides because nothing names it. A timesheet shows two entries for drafting a report and nobody reads that as the report being written twice. An email at nine in the evening attaching version four of a proposal looks like commitment, not waste.
Look in these places and you will usually find it:
- Proposals rebuilt from an older one, with the team CVs, the method section and the pricing table all retyped rather than pulled. See proposal and bid production.
- Client reports where the numbers are re-exported and re-pasted because the first pull was taken before month end closed. See client reporting.
- Compliance evidence reassembled from scratch at audit time, because nothing was captured as it happened. See compliance evidence.
- Onboarding packs where information is requested from a client who already supplied it to a different part of the firm.
- Internal summaries written for a partner because the underlying file is too long or too inconsistent to read.
That last one is worth dwelling on. When senior people cannot trust the record, juniors produce a second record for them. The firm then maintains two versions of the truth, and someone has to keep them aligned.
How do you put a number on it?
You do not need a study. You need a week and a small sample.
Pick four or five fee earners across different grades. Ask them to note, each day, anything they produced that already existed in some form, and roughly how long it took. Keep it to one line per item. Most people are surprised by their own list, because the pieces are individually small and never add up anywhere.
Then do the arithmetic the same way you would for any unbillable time: hours lost per person per week, multiplied by the number of fee earners it applies to, multiplied by your own charge-out rate, multiplied by 46 working weeks. The guide to calculating unbillable hours sets out the method and the assumptions to check, and the unbillable hours calculator runs the numbers.
Two cautions. Do not average across the whole firm if rework concentrates in one team, because the figure will look small and nothing will get fixed. And remember the number you produce is the floor, not the ceiling. It counts the hours. It does not count the work the firm turned down because those hours were gone, or the delay cost when a report lands late.
Why doesn't better discipline fix it?
The usual response is a drafting checklist, a style note and a reminder at the team meeting. These help for about a month. Then a deadline lands, someone copies an old document to save time, and the cycle restarts.
Discipline fails because it asks people to carry in their heads what the process should carry for them. The reviewer's standard, the correct precedent, the current figures and the house structure should arrive with the draft, not be remembered by whoever happens to pick the job up. That is a design question, not a behaviour question, which is the argument we make in a tool is not a process.
It is also why buying licences changed nothing in most firms. A general assistant can write prose quickly. It cannot know that your technical reports put limitations before conclusions, or that one reviewer will not accept a summary longer than a page.
What actually removes it?
Three things, in order.
- Agree the shape. Write down what a good version of the document contains, section by section, including what must never appear. Until that exists, every draft is a guess. How to document a process covers the method.
- Fix the sources. Decide where each fact comes from, so the figures, the scope and the client details are pulled rather than remembered. One source, one version.
- Produce the first draft from both. Built on the firm's own documents and language, in the tools the team already uses, so the author starts at a complete draft and spends their time on judgement. A named person still reviews and signs off anything that goes to a client.
The effect is that the correction loop shortens. Instead of three rounds of markup, there is one, and it is about substance.
Where should a firm start?
Start with the document your firm produces most often and argues about most. For an accountancy practice that is often the reporting pack; see accountancy firms. For engineering and surveying teams it is usually the technical report or the inspection record. For anyone chasing work, it is the bid.
Pick one. Count the rework in it for a week. If the number is uncomfortable, that is the process to rebuild first. The audit will point you at it in about three minutes if you would rather not guess.