Over-processing: effort the customer never asked anyone to spend

Specification that drifted tighter than the function needs

Ask a designer why a tolerance is what it is and the honest answer is often that it was copied from a previous drawing. Tight tolerances propagate: a feature is controlled closely, so an extra operation is needed, so a fixture is needed, so an inspection is needed. On the floor the same drift happens informally, when a machinist takes an additional finishing pass because a batch was once rejected and nobody told them the cause was different. Reviewing tolerances against function requires the design authority, and where the drawing belongs to a customer it requires their engineer, which is why it seldom happens.

Containment that became permanent

After an escape, the plant adds a check, a sort, an extra gauge or a second signature. That is correct at the time. What almost never happens is the review that removes it once the cause has been eliminated, because removing a check is an active decision with a visible downside and no visible upside. The remedy is procedural rather than analytical: record every added check with the reason, the person who introduced it and a point at which it will be reconsidered, then hold that review. Plants that skip this end up with an inspection routine assembled from a decade of forgotten incidents.

Paperwork and approvals are processing too

Data entered into one system and rekeyed into another, a job packet printed and then transcribed, a production record kept in parallel with a spreadsheet, approvals routed through people with no ability to judge the content. All of it consumes skilled time and introduces errors. The distinction worth drawing is between records a regulator, a customer or a standard genuinely requires and records the organisation invented for itself. The second category is usually larger, and much of it can be removed by whoever owns the procedure without any external conversation.

Why nobody volunteers to remove a step

The person who deletes a check owns the consequence if something escapes afterwards, while the saving is diffuse and belongs to nobody in particular. That asymmetry is the real obstacle, not analysis. It can be addressed by making removal a decision taken by a group rather than an individual, by requiring evidence that the original cause has been dealt with, and by trialling removal under observation before making it permanent. Senior support matters here: unless a manager states that removing an unnecessary step is expected work, no engineer will take the risk personally.

Testing whether a step is actually needed

Go to the next operation and to the customer's use of the feature, and find out what they do with the output. A finish nobody sees, a dimension nobody uses, a report nobody opens, a component cleaned before an operation that dirties it again. Where the answer is unclear, run a controlled trial: suspend the step for a defined quantity with the material identified and someone watching the downstream result. Evidence collected that way settles arguments that opinion cannot, and it gives the person accountable for the decision something concrete to stand behind.

Frequently asked questions

How do we tell over-processing from genuine quality work?

Trace the requirement back to a source. If a customer specification, a regulation, a standard the plant is certified against or a demonstrable functional need calls for it, the work is necessary even though it adds cost. If the only justification is that it has always been done, that a previous manager wanted it, or that it makes the part look better than it needs to, the step is a candidate for removal. Write the source next to each requirement and the ones with no source stand out immediately.

Our operators add extra finishing work on their own initiative. Is that a problem?

It is, for two reasons beyond cost. It means the standard does not describe what is actually done, so cycle times and capacity figures are wrong, and it means the result varies with who is on shift. Find out what they are protecting against, since there is usually a real reason such as a fixture that drifts or a gauge nobody trusts. Fix that, then rewrite the standard, rather than simply instructing people to stop.

Can removing a step invalidate a customer approval?

It can, particularly where the process was submitted and approved as part of a part approval package, or where the product is regulated. In those cases the step cannot be dropped unilaterally regardless of how unnecessary it looks, and the route is a formal change request to the customer or the relevant authority. Identify which steps carry that status before starting, so the team spends its effort on the large remainder that the plant is free to change.