Revenue diversification — the risk that never shows up in P/E

Revenue diversification measures how spread out a company's revenue is across customer relationships, products and geographies — most often expressed as HHI (the sum of squared revenue shares) or as the largest customer's share of revenue. A company where the largest customer relationship accounts for 40 % of revenue carries a risk that the P/E ratio is completely blind to: two companies can show identical earnings and an identical multiple and still have entirely different probability distributions for the future. It is that risk — the one that does not show up in P/E — that revenue diversification captures.

The formula — two measures of the same exposure

HHI = Σ (share_i × 100)²

where share_i is the revenue share in percent for each customer relationship or segment. The scale: eight equal shares of 12,5 % give 8 × 12,5² = 1 250; a single customer relationship at 100 % gives 10 000. Competition law uses the same measure for a mirrored purpose — measuring seller concentration — and from there come the thresholds that also hold analytically: below 1 500 low concentration, 1 500–2 500 moderate, above 2 500 high. Turn the measure toward the company itself and you have a scale for how much of the revenue stream rests on few legs.

The simpler complement:

Top-customer share = Largest customer's revenue ÷ Net revenue × 100

HHI captures the whole distribution; the top-customer share answers the bluntest question: how much disappears if a single relationship breaks? Use both — they answer different questions.

The worked example: same P/E, entirely different underlying risk

Two hypothetical companies, both with 150 Mkr in earnings and both traded at P/E 15 — in other words, the market capitalization is 2 250 Mkr each:

• Company A — subcontractor to the automotive industry. Revenue shares: 40, 20, 15, 10, 8 and 7 %. HHI = 40² + 20² + 15² + 10² + 8² + 7² = 1 600 + 400 + 225 + 100 + 64 + 49 = 2 438. Top-customer share: 40 %.
• Company B — consumer brand. Eight revenue segments, none above 12,5 %. HHI = 8 × 12,5² = 1 250. Top-customer share: 12,5 %.

Same earnings. Same P/E. But if Company A loses its largest customer relationship — 40 % of revenue, and since the volumes carried the operation probably the greater part of operating profit — next year's numbers look entirely different. For Company B the same event is an absorbable shove: 12,5 % of revenue disappears and the business keeps running. The multiple was the same — the risk was not. A P/E discounts this year's earnings; concentration affects the probability distribution of all future earnings. That is why revenue diversification is not a valuation variable but a risk variable — and why it is called V03 in the AKM1 model, where it is weighed next to V01 and V02 but always read as a risk filter on the entire growth block.

The three dimensions — and how they differ

The customer dimension is only one of three:

• Customer: how many independent payers stand behind the revenue?
• Product: how many revenue-driving products are there? A company can have a large collective of buyers and still rest its entire revenue on a single platform that a technical transition can knock out.
• Geography and currency: the same customer base spread across several countries spreads both business cycle risk and currency risk; a single home market concentrates both.

A complete reading goes through all three. The raw material is the segment reporting in the notes to the accounts of the annual report — with the important proviso that the companies choose the segmentation themselves, usually by product or geography and rarely by counterparty. The true customer concentration lies one level deeper, in the note on significant counterparties (see pitfall 1).

Pitfall 1: segments are not counterparties

The reporting requirement (IFRS 8) covers operating segments — product or geographic areas — while individual counterparties only have to be reported if they pass the 10 % threshold for revenue or assets. The consequence: the reported top-customer share is a lower bound, not an exact level. A segment labelled Europe 55 % of revenue can at its core rest on a single distributor. The level beneath the segment notes is the counterparty list; when it does not exist, that is a gap in the material, not proof of diversification.

Pitfall 2: the number of heads is not the same as independence

Ten customer relationships that all sell into the same end market — ten subcontractors to the automotive industry, or two formally separate counterparties in the same group — are correlated. HHI counts heads, not co-movement. In a downturn, correlated revenue falls at the same time and the diversification turns out to be an illusion. The better question: if the customers' own customer market invests 20 % less next year — how large a share of the company's revenue is affected then? It is that share that is the true concentration.

Pitfall 3: diversification that has been bought

Acquisitions in new business areas lower HHI on paper but stack integration risk and the risk of a splintered management focus on top of the one that disappeared. Peter Lynch called the phenomenon diworsification — diversification that makes things worse. By all means count revenue shares by segment as they look acquired, but read a falling HHI with skepticism if it comes only from acquisitions: organically grown spread is the kind that holds when the business cycle turns.

The industry interpretation — what is normal

Descriptively, with no yardstick beyond the comparison group: project-based industries such as construction, contracting and systems delivery live naturally with high concentration — a few large clients carry the year's revenue, and an HHI above 2 500 is the norm rather than the exception. Retail and consumer goods sit at the other end: thousands of buyers per day give an HHI far below 1 000 — there it is instead product and channel risk that carries the analysis. B2b suppliers with a handful of industrial counterparties fall in between, and that is exactly where the correlation question (pitfall 2) makes the biggest difference. The point is not that an HHI is good or bad in absolute terms, but that the threshold should be read against the industry's structure: an HHI of 2 400 is high for a consumer company and completely normal for a contracting company. In the AKM1 model, V03 is therefore always scored against industry type — the same logic that makes ROE (V09) interpreted differently in banking and software.

The exercise: measure concentration in a real annual report

1. Open the latest annual report of a company you follow and find the segment reporting and the note on significant customer concentration. 2. Build a table of revenue shares — by segment as a minimum, by counterparty if the notes to the accounts go deep enough. 3. Calculate HHI and top-customer share for the two most recent years. Is concentration rising or falling, and why? 4. Enter the shares in the calculator and see the placement against industry-typical levels.

A quarter of an hour, one table — and you see the risk that the multiple never shows.

Next steps

• Continue in the course Revenue diversification (V03): six chapters — chapter after chapter — on HHI, segment reading and concentration risk.
• See how V03 works as a risk filter in the Curriculum — growth resting on too few legs is counted differently.
• Add the whole picture in the portfolio builder: diversification at company level and at portfolio level are two different questions.

FAQ

What is revenue diversification, and why doesn't it show up in the P/E?

Revenue diversification measures how spread out a company's revenues are across customers, products and geographies — usually expressed as HHI or the largest customer's share of revenue. The P/E discounts this year's earnings but says nothing about the probability distribution of future earnings: two companies with identical profit and identical multiple can carry entirely different risk if one of them rests 40 percent of its revenues on a single customer relationship. That is why revenue diversification is a risk variable (V03), not a valuation variable.

How do I calculate a company's HHI?

Kvadrera varje intäktsandel uttryckt i procent och summera: HHI = Σ (andel_i × 100)². Åtta lika stora andelar om 12,5 % ger 8 × 12,5² = 1 250, medan en enda kundrelation på 100 % ger 10 000. Trösklarna är under 1 500 låg koncentration, 1 500–2 500 måttlig och över 2 500 hög. Kalkylatorn räknar om andelarna och visar placeringen mot branschtypiska nivåer.

What is a normal HHI level — and what counts as high concentration?

Det beror på branschens struktur. Projektbaserade branscher som bygg, entreprenad och systemleveranser lever naturligt med HHI över 2 500, medan detaljhandel och konsumentvaror ofta ligger under 1 000 och i stället bärs av produkt- och kanalrisk. B2b-leverantörer hamnar mittemellan. Samma HHI kan alltså vara normal för ett entreprenadbolag och högt för ett konsumentbolag — i AKM1-modellen poängsätts därför V03 mot branschtyp.

Why isn't counting the number of customers enough?

A common misconception is that many customers automatically mean diversification. HHI counts heads, not co-movement: ten customers all selling into the same end market are correlated and fall at the same time in a downturn. Moreover, reported segments are not counterparties — a segment labelled Europe 55 percent can rest internally on a single distributor. The better question: if the customers' own market buys 20 percent less next year, how large a share of revenues is affected?

This is educational financial analysis, not investment advice.