Revenue by Segment: Five Divisions, One Pays the Bills

"Revenue grew 8%" is not information. It is a sum. Only revenue by segment, by region and by type tells you what actually happened: a strong year in hardware, flat hardware rescued by subscriptions, or one market shrinking underneath a positive headline. Diversification lives in that split, not in the number of divisions on a slide. Five segments where one earns most of the profit are one business and four accessories.

Where the segment note sits#

The segment note is one of the few places where a company takes its own revenue apart and signs its name under the result. In a US 10-K it sits among the notes at the back; in Apple's report for the fiscal year ended 27 September 2025 it is Note 13, "Segment Information and Geographic Data". Under IFRS you look in the notes for operating segments. The annual report carries a dozen or more notes, and this is the one to open before the income statement, because without it the income statement is a column of numbers with no address.

Copy four columns per segment into a table of your own: revenue for the last three years, segment profit, every cost the company broke out separately, and one sentence on what the segment actually sells. Three years, because two points make a line, not a trend. If the table does not fit on one page, the company is either genuinely complicated or reporting in a way that discourages arithmetic.

Except that this map has an author.

Segments are not a category imposed from outside. The standard asks a company to present the business the way the person who allocates its resources sees it, so the split reflects how the thing is managed rather than how the market thinks about it. Apple says so plainly in its 10-K for fiscal 2025: "The Company manages its business primarily on a geographic basis". So its reportable segments are the Americas, Europe, Greater China, Japan and the Rest of Asia Pacific, not iPhone and Services. The product split lives in a separate note, the one on revenue.

Three numbers per segment#

Pull three numbers from every segment, and none of them is enough on its own: share of revenue, year-on-year growth, and share of profit. One says how wide the stream is. The second says whether it is rising. The third settles which one actually keeps the company afloat.

One question comes before all three, and it is narrower than the usual one: which segment paid for this growth? Not which grew fastest in percentage terms, because the fastest can contribute less money than a large segment that barely moved. What matters is how many dollars of profit arrived, and which channel carried them.

Say a company has two segments. The first earns 800m of revenue and 40m of operating profit, the second 200m of revenue and 60m of profit. That is a billion of revenue and 100m of profit, which means the segment responsible for a fifth of sales delivers three fifths of the result. If total revenue now grows 8%, the question is not "by how much" but "in which channel". Eighty million added by the first segment lifts profit by 4m. The same 80m in the second lifts it by 24m.

Same headline, six times the effect.

Read revenue by segment from the profit column#

A real company shows the same arithmetic at scale. Apple sold $109.2bn of Services in fiscal 2025 against $416.2bn of total net sales, which makes Services roughly a quarter of the top line. In gross margin the same item weighs $82.3bn out of $195.2bn, which is more than two fifths. One pair of numbers from the filing explains the gap: gross margin ran at 75.4% on Services and 36.8% on Products.

Which means a dollar of Services revenue leaves more than twice as much in the company as a dollar of hardware revenue. A quarter of sales funds close to half of everything that happens below the gross margin line: research, stores, the dividend, the buyback. A segment that looks like a sideline in the revenue table can be holding up the rest of the group.

So I read the segment note from the profit column, not the revenue column.

To watch the same journey at the level of the whole business, follow what happens to one dollar before it becomes profit.

That same pair of shares catches the thing that erodes margin quietly: mix shift. When the cheaper, lower-margin segment grows faster than the expensive one, total revenue keeps looking encouraging for several quarters while group margin slides with no dramatic cause attached. Management will later call it a demanding market environment. The segment note calls it by its name, provided somebody compares last year's shares with this year's.

Segments finally show their costs#

Until recently this note ended sooner than the questions started: segment revenue, one line of profit, and everything in between stayed inside the company. Filings for fiscal 2025 are different, because ASU 2023-07 requires disclosure of the segment expenses that are significant and regularly reported to the chief operating decision maker. Apple describes adopting it in its accounting-policies note, and its segment table now shows cost of sales and selling costs separately for each region.

For a reader one thing changes, and it is the important one: you compute the segment margin yourself instead of taking a single supplied number on trust.

There is a condition attached that is easy to forget. Research and development and general costs stay unallocated at Apple, and the fiscal 2025 filing carries them on a single line of $42.6bn. That is exactly the amount by which the five segment profits exceed group operating income, and more than two thirds of every cost the company booked below the gross margin line. So a margin computed from this note is a margin before corporate costs. Compare it with another segment of the same company, or with the same segment a year earlier, never with a competitor's operating margin.

Sales geography is not operations geography#

Regions are where a total keeps quiet about the thing that matters. Fiscal 2025 closed at $416.2bn of net sales against $391.0bn a year earlier, growth of 6%. Underneath that total, Greater China fell 4% to $64.4bn, its second decline in a row after an 8% drop the year before. Japan grew 15% over the same period and Europe 10%. The company's third-largest market has been shrinking for two years, and the total never once said so.

I do not know whether that segment returns to growth next year. I do know the 6% headline does not answer the question and never will.

There is a second geography that this note does not contain. Where does the company manufacture? The Americas, Europe and Japan tell you only where it sells, and say nothing about where components come from or which government can add a tariff or a levy tomorrow. That lives in the risk section, next to the description of supplier concentration, which is why the segment note is read alongside the risks a company was obliged to disclose. A company with half its sales in Europe and all its production in one Asian city has five markets in the table and one bottleneck.

Separating the two maps has a practical consequence for currency. A segment that sells in yen and buys components in dollars hands part of its margin to the exchange rate, while looking identical in the revenue table to a segment settled in one currency on both sides. So every region deserves two notes: what the company invoices in, and what it pays its bills in. If those two currencies differ, a good year for that segment is partly a currency move rather than demand.

The quarter that means nothing on its own#

Segments are read to the rhythm the company's calendar imposes, not the reader's. Apple warns in the fiscal 2025 filing: "The Company has historically experienced higher net sales in its first quarter compared to other quarters in its fiscal year due in part to seasonal holiday demand". Which means a sales decline between its first and second fiscal quarters is a calendar norm rather than an event, and comparing one quarter with the previous one mixes a change in the business with the rhythm of the holidays.

Two habits settle it. Compare a quarter with the same quarter a year earlier. Check the length of the period in the accounting-policies note, because Apple's fiscal year ends on the last Saturday of September and gains an extra week every five or six years: 2023 ran 53 weeks, 2024 and 2025 ran 52 each. One extra week of selling can manufacture growth that is not there.

Not every dollar is the same species#

Two segments with identical revenue and identical margins can be worth different money, because they differ in how far ahead you can see the stream. A subscription is contracted. Revenue from habit repeats, but nothing guarantees it. Project work has to be won again every year, and cyclical revenue depends on customer capital spending, commodity prices or interest rates, and can evaporate in two quarters.

With subscriptions the sales line says too little, because it blends what was won with what was kept. Net Revenue Retention separates them: how much revenue is left from the same group of customers a year later, after churn, downgrades and expansion. Above 100% means existing customers pay more than they did a year ago even if the company signed nobody new.

Attach a date to revenue quality. How far ahead can you see this stream? A contract expires in a specific year, a backlog converts into cash on terms somebody wrote down, and a subscription either renews or does not. A separate question is what the growth itself is made of – price, volume, currency or an acquisition – and that has its own piece on what revenue growth is made of.

When diversification is only a brochure#

The word "diversified" appears in company materials more often than almost any other and is almost never counted. The test is cheap and takes a minute: write down each segment's share of profit and check whether the first one eats most of it. Five divisions where one delivers four fifths of the result are one business with four expensive hobbies.

The second test is sharper, because it asks about the source rather than the number of channels. Apple has five reportable segments and all five sell largely the same thing: iPhone accounted for $209.6bn of the $416.2bn of fiscal 2025 net sales, which is more than half the group. The regional table looks like spread risk, yet a single question about demand for one phone touches all five rows at once.

I do not believe in diversification that fails to show up in the profit column and falls apart at one question: what happens when that one stream runs dry.

Concentration by itself is not an accusation. A defence manufacturer depends on government budgets because that is the job, and nobody sensible expects five even segments from it. The question is whether the company kept its bargaining power with that one counterparty, and that question is asked separately, at the note on customer concentration. Bad concentration is not the large kind; it is the kind management says does not exist.

The most misleading brochure of all appears when a new segment arrived by acquisition. In the table it looks like a second channel. Often it is a ditch dug alongside the river, running its own water for years: its own costs, its own debt, its own managers, half of whom are leaving. Bought revenue is real revenue, right up to the point where the two sides start arguing over it. So check whether the acquired segment earns anything close to the promised margin, and whether the company keeps reporting it separately for more than a year, because an acquisition folded into a shared bucket after twelve months can no longer be measured against the promise.

When a company redraws its own map#

Reorganising segment reporting is the cheapest way to make an inconvenient trend stop being visible, and it is usually not done for that purpose, which does not excuse you from care. Rebucketing breaks the continuity of the series, and new categories can blend businesses with very different margins. The company publishes recast comparatives alongside the change, and those are what you rebuild the history from, not the old reports.

Three steps when you see one:

1. Rebuild the series from the recast figures for prior years, not from the old presentation.
2. Write down what became clearer after the change, and what stopped being visible on its own.
3. Note a question for the next earnings call about the segment that disappeared into a bigger bucket.

At Taufolio this work sits in two places. The Full report takes a company apart into segments and geographic exposure, with a link to the filing behind each number, and costs the most credits of any report. The Monthly summary answers the narrower question: which stream changed since last time, and whether the change touched the segment that pays the bills. What the breakdown looks like on a finished document is on the product page.

A rule I have not given up on in years: until you can name the segment that pays the bills and say what could dry it up, you do not know the company yet. You know its total.