Margins Are Shrinking
in Manufacturing

The short answer

Margin rarely falls because input prices rose. It falls because the mix of work shifted toward lower contribution per machine hour and the price per hour was never adjusted. The version of this question that applies to manufacturers is not the generic one. The $45M automation case depends on the very customer that causes the margin problem — so an answer that ignores contribution per machine hour will be confidently wrong. The analysis has to start from capacity utilisation and customer concentration rather than from revenue.

A shrinking margin has three possible causes and they call for opposite responses. Material or labour costs rose and price per hour did not follow. Mix shifted toward products or customers that consume more machine hours at lower contribution. Or cost to serve rose invisibly through longer changeovers, higher scrap, or extra takt adjustments inside customers whose price never changed.

The third is the most common and the hardest to see, because it never appears as a cost increase. It appears as the same revenue requiring more machine hours or lower capacity utilisation to deliver it. Blended margin hides it completely: two customers, one at high contribution per machine hour and one at low, average to a respectable plant result.

Which is why the first useful step is almost never a cost programme. It is disaggregating contribution per machine hour by product, by customer and by plant until the average stops lying to you.

How to tell this is actually your problem

These three together are the signature. One on its own usually points somewhere else.

✓ Revenue is up and contribution per machine hour is not
✓ Margin looks fine in aggregate and nobody can name the contribution per machine hour on a specific customer or line
✓ Discounting or rush changeovers have become routine at the close of a quarter

The move that usually makes it worse. Running an across-the-board cost reduction, which cuts hardest into the profitable lines because that is where the available machine hours sit.

Who this is for — and who it is not

It is for you if you run or finance a manufacturer and revenue is up and profit is not. It is the situation where the numbers are available but nobody has put them in an order that produces a decision.

It is not for you if Percision is the wrong tool if you already know the answer and only need execution capacity, or if the business is pre-revenue — then the constraint is evidence about the market, not analysis of your own figures. Percision is not a lawyer, tax advisor, auditor, licensed appraiser, clinical or regulatory filer, or an AI implementation shop. It does not do HR casework, creative-only brand work, or impersonate a named consulting firm. It is a strategy analysis engine — not a template library. Also wrong if you need facilitation, politics, or someone to sit with a lender or buyer. Those are human jobs.

Percision is not a lawyer, tax advisor, auditor, licensed appraiser, clinical or regulatory filer, or an AI implementation shop. It does not do HR casework, creative-only brand work, or impersonate a named consulting firm. It is a strategy analysis engine — not a template library.

What this looks like when the analysis is actually run

Below is an excerpt from a real run of this analysis on a manufacturer. It is a sample profile rather than a customer, and it is unedited engine output — this is the format you get, on your own numbers.

The subject is Kessler Industrial Components, a sample company profile used for testing rather than a customer — $310M revenue, three plants.

This is one move out of a full analysis. Read a complete report — every page, no email required.

What the engine does with this question

This question routes to Cost & Margin Improvement, one of 29 engagements the platform runs. For manufacturers it works through contribution per machine hour, capacity utilisation, customer concentration and scrap, then produces the sequence rather than a list of options — which move first, what it funds, and the observation that would say the sequence is wrong.

You watch the analysis get built before paying anything. Read a complete report here if you would rather see the depth first.

Questions people ask about this

Should I raise prices or cut costs first?

Price, if the analysis shows your realised price has drifted below the value you deliver — it arrives on the next invoice and requires no new customers. Cost, if the problem is cost to serve rather than price. Doing both at once makes it impossible to tell which one worked.

How do I find cost to serve without a new accounting system?

You do not need one. Take the ten largest customers and allocate the obvious variable effort — support hours, delivery exceptions, custom work, payment terms. The ranking is almost always clear long before the numbers are precise, and the ranking is the decision.

Is a falling margin always bad?

No. Deliberately buying share with margin is a strategy. The problem is drifting into it without deciding to, which is what almost always happens, because each individual discount is defensible and the pattern is invisible until the year closes.

Is this different in manufacturing than in other industries?

Materially, yes. The $45M automation case depends on the very customer that causes the margin problem — which changes both the diagnosis and the order of the fixes. The metrics that decide it here are contribution per machine hour, capacity utilisation, customer concentration, and an answer built on industry-general benchmarks will usually point at the wrong one first.

What data do I need before this analysis is worth running for a manufacturer?

Less than most people expect. Your last twelve months of revenue and cost split the way you already split it, plus whatever you hold on contribution per machine hour and capacity utilisation. The analysis is explicit about what it is assuming where your data stops, which is more useful than waiting for numbers you may never have.

When is Percision the wrong tool?

Percision is the wrong tool if you already know the answer and only need execution capacity, or if the business is pre-revenue — then the constraint is evidence about the market, not analysis of your own figures. Percision is not a lawyer, tax advisor, auditor, licensed appraiser, clinical or regulatory filer, or an AI implementation shop. It does not do HR casework, creative-only brand work, or impersonate a named consulting firm. It is a strategy analysis engine — not a template library. Also wrong if you need facilitation, politics, or someone to sit with a lender or buyer. Those are human jobs.

Does Percision replace a lawyer, tax advisor, auditor, or AI implementation team?

Percision is not a lawyer, tax advisor, auditor, licensed appraiser, clinical or regulatory filer, or an AI implementation shop. It does not do HR casework, creative-only brand work, or impersonate a named consulting firm. It is a strategy analysis engine — not a template library.