Rework: the maintenance hours you spend undoing your own mistakes

A pump comes back from the repair bench and goes in on Monday. By Thursday it has failed again. The crew groan but say little beyond "it just wasn't right". A fresh work order is raised, parts are pulled, and four more hours disappear before anyone asks the harder question: why did it fail so soon after the last repair?

Without a way to flag and count that kind of repeat work, the cost stays invisible. It blends quietly into the monthly labour total, and the pattern that caused it goes unbroken, ready to happen again.

What it actually measures

Rework is corrective work on equipment that failed early because of a maintenance error, an operating error, or a material quality problem, expressed as a share of your total maintenance labour hours. Some call the repeat visit a callback, and the idea is the same: if you were paid to do the job, you would not bill again for it, because it should have been right the first time. Most definitions put a clock on it, counting a failure as rework if it comes back within a set window of the last job, anywhere from twenty-four hours for an emergency to thirty days after a planned repair, and the exact window matters less than applying it consistently.

What sets rework apart from ordinary breakdown maintenance is the cause. A normal repair traces back to honest wear; rework traces back to something that went wrong in the previous intervention.

How to work it out

Rework (%) = Corrective hours identified as rework / Total maintenance labour hours × 100

In a month the team works 1,000 hours, of which 40 are spent on jobs traced back to a previous maintenance or operating error.

Rework (%) = (40 / 1,000) × 100 = 4%

Four of every hundred maintenance hours that month went on undoing errors rather than adding value. It sounds small, but it is pure waste, and the only acceptable direction for it is down.

What good looks like

A widely cited benchmark puts good performance somewhere under 2 to 3%, but the guiding principle matters more than the precise figure: rework should be as close to zero as you can get it, and the trend should always point downward. A consistently low number is quiet proof that your procedures, your training and your materials quality are all doing their job.

Why the work was wrong in the first place

Here is the uncomfortable truth behind rework: a great deal of equipment failure is self-inflicted. Studies of industrial plants have put the share of self-induced failures as high as seven in ten, and a large slice of those trace back to people simply not knowing the basics, with one skills assessment finding that the overwhelming majority of technicians lacked the full fundamentals of their craft. The maintenance can be the defect. When one plant inspected the bearings it was replacing on a fixed schedule, only a small fraction were genuinely near failure; most of the rest showed only minor damage from poor lubrication, alignment or installation, harm the maintenance itself had introduced. As the saying in the trade goes, low-reliability people end up servicing highly reliable equipment.

That reframes rework from a scolding into a to-do list, and the fix is well worn. Write detailed procedures and follow them. Train people in why thorough work matters, not just how. Resist the urge to over-fix: if it is working, leave it, and keep a sense of proportion about what to put right on the spot versus what to schedule separately. And run a full operational test after the job, before you call it done, because a surprising share of failures show up in the first hours and days after a repair, exactly where a proper test would have caught them.

Where it can mislead you

• You need a separate work order to capture it. Fold the repeat job back onto the original work order and the rework vanishes from view, taking the metric's whole value with it.
• The early failure must be genuinely traceable to a previous activity. Not every premature failure is rework, and forcing the label on honest wear muddies the signal.
• Because every organisation defines rework a little differently, it is a poor metric for comparing one plant against another; its real power is watching your own trend over time.
• Do not let it become a stick for the crew. A callback is often caused upstream, by a hasty patch ordered under pressure or an inferior spare part, not by the person who did the work. Break the figure down by craft, crew, procedure or supplier to land the investigation in the right place rather than on the nearest technician.

Rework is one of the few numbers that is almost entirely within your power to improve, because every hour of it came from something that was done, not from the equipment's nature. Measured honestly, it turns a nagging sense that "we keep going back to that one" into a percentage you can investigate, shrink, and one day all but eliminate, and because it is so closely tied to skill, driving it down tends to lift everything else along with it.