Overtime rose and completed orders did not. That is a signal—not an explanation.
Overtime rose by 18%. Completed orders did not.
One manager wants more people. Another wants new scheduling software. A consultant recognises a training opportunity. Everybody has a cure. Nobody has yet explained what happened.
The same result could come from a harder product mix, rework, missing material, equipment downtime, absenteeism, poor work release, a customer promise the operation cannot meet or completed work sitting somewhere short of invoicing.
Those causes do not share one sensible cure.
Before management spends money, it needs to know which explanation survives contact with the evidence.
A symptom is only the starting signal
“Overtime is up” describes something visible. It does not tell us where the loss entered the work, how far it travelled or what decision would change it.
The first job is to state the concern and the decision the analysis must support. Then define the business perimeter: which operation, period, locations, products and connected activities belong in the question.
Only after that should we choose the comparisons.
Output against input can be useful, but there is no magical output-to-input ratio that diagnoses a business. A ratio shows that a relationship changed. Diagnosis begins when we explain why.
If completed orders stayed flat, did the number of orders change, or did their complexity? Did work start but fail to finish? Did finished work wait for inspection, documentation or invoicing?
If paid hours rose, where did they go? Into saleable work, rework, waiting, changeovers, breakdowns or extra supervision?
If material cost rose, was the cause price, product mix, scrap, poor yield, urgent purchasing or stock that never became saleable output?
These are different questions. Lumping them into “productivity” merely gives the uncertainty a respectable name.
Test explanations against what should be visible
For every plausible explanation, ask four things:
- If this is true, what should we see in the records or in a recent real case?
- What evidence would weaken or contradict it?
- When did the change begin, and what else changed then?
- Who owns the evidence, and is it reliable, estimated or disputed?
Suppose management believes overtime rose because demand became more complex. If that is true, standard or expected hours per order should rise with the product mix. If first-pass yield fell at the same time, rework becomes a competing explanation. If late material arrivals cluster around the overtime peaks, the problem may begin in supply rather than labour capacity.
This is why one recent example followed from start to finish is usually worth more than ten explanations beginning with “normally”. Records show the pattern. People explain the workarounds, decisions and missing facts behind it. Neither source is sufficient alone.
In a remote diagnostic, there is another discipline: state what cannot be concluded without another interview, better records or seeing the operation. Distance is a boundary, not a reason to make the language more confident.
Do not manufacture a saving
A diagnosis should quantify only what the evidence and management authority can support.
Overtime, rework, scrap, urgent freight and customer credits may be direct impactable costs. Fixed overhead is not suddenly a saving because a spreadsheet divided it by fewer units. Freed labour hours are capacity, not cash, until the organisation uses them for more or better saleable output, genuinely removes or avoids the cost, or combines the two.
The three practical routes are still clear:
- produce more or better saleable output with the same resources;
- maintain the output with fewer resources actually consumed; or
- combine the two.
Known amounts, estimates and sensible ranges must remain separate. Connected findings must not be counted twice. The same rework hours cannot become one saving under labour, another under overtime and a third under delayed delivery merely because three columns are available.
The result must support a decision
A paid diagnostic should stand on its own even if no implementation assignment follows.
For each material finding, management should receive:
- the evidence supporting it;
- the assumptions and uncertainty still attached to it;
- the likely operational and financial effect;
- what the evidence does not support; and
- the next decision, action or fact that must be obtained.
Sometimes the right decision is to act with the company's own people. Sometimes specialist or on-site work is justified. Sometimes the suspected leak is too small to deserve attention. A useful diagnosis can say that too.
The same discipline applies when preparing for ISO 9001:2026, which is currently under publication. Rewriting documents before checking how decisions, failures and handovers work in practice is another cure looking for a diagnosis.
Before approving an expensive solution, ask: What exactly have we observed? Which explanations remain credible? What evidence supports or contradicts each one? Which part can management actually change? What decision will the answer enable?