Five traps that look like a migration

The costly errors in this kind of analysis are not the businesses that obviously sell a commodity. Those are easy. The costly errors are the arrangements that produce precisely the pattern a migration produces — margin stepping up, mix apparently improving, a multiple that still reflects the old business — while the economics underneath are unchanged or worse. Each of the five below is common, each is entirely legal and usually well disclosed, and each is invisible if you read group-level numbers without restating them.

One: divestment arithmetic

Group margin is a weighted average. Sell the least profitable division and the average rises the next morning, with no operational change anywhere in what remains. Reported revenue falls, reported margin improves, and a chart of group profitability shows a clean inflection at the transaction date.

The tell is the shape. Genuine mix shift is gradual, because it comes from one line growing faster than another over years. Divestment arithmetic is a step, and the step lands in the period of the disposal. The restatement that settles it is to compute margin on the continuing perimeter for every year in the window — same businesses, before and after. If the improvement survives that, something real happened. If it disappears, what happened was subtraction.

This one is worth taking seriously rather than dismissing, because portfolio pruning can be genuinely good management. A company that exits a structurally poor business has improved its future, just not through the mechanism a migration thesis assumes. The distinction matters because subtraction runs out: you can only sell the worst division once.

Two: bought specialty

A company acquires a differentiated business and the group’s mix instantly looks better. Higher-value revenue as a share of the total jumps, margin rises, and the strategy narrative writes itself. Nothing here is illegitimate — purchasing capability is a normal way to change what a company is — but the evidence required to believe it is different from the organic case.

Three things need checking. What was paid, because a specialty business bought at a full price transfers the economics to the seller and leaves the buyer with the goodwill. Whether the advantage survives ownership, since differentiation often rests on people, customer relationships and process knowledge that integration disturbs. And how the margin is being presented: acquisitions create intangible amortisation, and a margin quoted before that charge is not comparable to the pre-deal margin it is being compared against.

The organic version of a climb is slower and harder to see, but it has one property the acquired version lacks — the company demonstrably knows how to build the capability, which means it can do it again.

Three: redrawn segments

Reporting segments follow internal management structure, so when a company reorganises how it reviews results, the external disclosure changes with it. Sometimes this reflects a genuine change in how the business is run. Sometimes a struggling line is folded into a healthier one and stops being separately visible.

Either way, a mix trend computed across a segment redefinition is not a trend. The check is mechanical: read the segment footnote’s basis-of-presentation language, establish whether prior periods were restated onto the new definitions, and if they were not, rebuild the comparison yourself or refuse to draw the conclusion. A redefinition that coincides with the start of a transformation narrative is not proof of anything on its own, but it does mean every subsequent comparison needs doing by hand.

Four: one customer, briefly

A company wins a programme with a large customer, the higher-value revenue line steps up, and margin follows. This looks like a qualification-won, moat-building climb — and sometimes it is exactly that. But revenue concentrated in a single customer has a different risk profile from revenue spread across many, and the concentration is frequently greater in the new business than in the old one it is replacing.

What separates the two cases is the switching cost. If the component has been designed into the customer’s product, qualified through a long approval process, and is painful to second-source, the concentration is the flip side of a genuine moat. If it is a supply arrangement won on price and re-tendered periodically, then the company has swapped diversified commodity revenue for concentrated commodity revenue and improved nothing but the near-term margin. Disclosure requirements around significant customers, and any long-term agreement filed as an exhibit, are where that distinction becomes readable.

Five: capitalised costs

Costs that are capitalised rather than expensed leave the income statement and reappear as depreciation or amortisation spread over later years. Margin rises immediately; cash generation does not. A company in the middle of a large build can show improving reported profitability while its free cash flow goes sideways or negative, and the reported improvement is at least partly a timing difference.

This is why the cash flow statement is the referee. Compare operating margin with cash conversion over the same five-year window: a genuine climb usually shows both improving, with cash lagging investment for a period and then catching up. A gap that persistently widens — rising reported margin, static or deteriorating cash — is a signal to work out exactly which costs stopped being expensed and when.

The five at a glance

Trap	What it mimics	The restatement that exposes it
Divestment arithmetic	A margin inflection	Margin on the continuing perimeter, all years
Bought specialty	A mix shift up the chain	Organic mix excluding acquisitions; margin after intangible amortisation
Redrawn segments	A rising high-value share	Comparison on a single consistent segment definition
One customer	A qualification-won moat	Concentration disclosure and contract term against switching cost
Capitalised costs	Operating improvement	Operating margin against cash conversion, same window

The common structure is worth noting: in every case the reported figures are accurate and the inference is wrong. These are not accounting scandals. They are ordinary disclosures being read at the wrong level of aggregation, which is why the defence is arithmetic rather than suspicion.

What this framework misses

It catches arrangements that flatter the numbers. It does not catch a company that genuinely climbs into a bad market, executes a real transition that arrives late, or builds a differentiated position that a competitor then copies. Those are failures of strategy and timing rather than of presentation, and no amount of restatement finds them.

It also cuts both ways. Each of these mechanisms can run in reverse: an acquisition can obscure a genuine organic climb, a segment redefinition can hide a line that is genuinely improving, and heavy expensing during a build can make a real transition look worse than it is. The point is not that these patterns are evidence of a trap, but that they make the group numbers uninformative until the like-for-like version has been computed.

The bottom line

Before believing any group-level improvement, restate it: same segments, same definitions, same perimeter, several years, and cash alongside profit. Most of the traps above dissolve under that treatment, and the ones that survive it are worth a closer look. That restatement is the unglamorous half of the structural-versus-cyclical test, it is what the Analysis Bridge is built to run at scale, and it is why the research book publishes the names the framework passes on alongside the ones it does not.