Chapter 2

Who Sets the Unit Price

The whole of the payroll-heavy income statement the reader now holds resolves to a single question about one thing: the price of an hour of Accenture's labour, and whether it holds while the technology the company sells makes that hour cheaper to produce. A reader would reach, reflexively, for the instrument that measures it — a rate card, a realization index, a price-versus-volume split. Accenture states in its own 10-K that it does not have one.

The instrument the company says it lacks

In the section of the FY2025 Form 10-K where a manufacturer would decompose revenue into price and units, Accenture writes the opposite: "we cannot measure how much of our revenue growth in a particular period is attributable to changes in price or volume. Management does not track standard measures of unit or rate volume." [1] Each contract is a customized mix of services, the filing explains, so no standard comparability measure applies. The single most important operating variable in a labour business is, by management's own account, unobservable from the outside.

The word the company does use — "pricing" — is not a price. Asked on the Q2 FY2025 call whether the pricing environment was stabilizing, the CFO answered and then defined the term: "pricing is the contract profitability or margin on the work that we sell." [2] That reframes the entire disclosed pricing arc. When the same analyst noted that "I looked back the last seven quarters, you called pricing as being down a bit," [3] and management later described pricing improving through Q3 FY2025 and into FY2026, both were talking about contract margin, not the rate charged per unit of work. The seven-quarters-down-then-recovering series that reads like a price index is a margin commentary. There is no rate series, no realization figure, no win-rate and no churn rate anywhere in the covered filings and calls. The absence is not a gap in this chapter; it is the chapter's subject.

Two proxies, both against a ceiling

With the direct instrument gone, two proxies remain, and both are pinned. The first is volume, for which the only observable series is headcount. In FY2025 revenue grew 7.4% while the year-end workforce grew about 0.6%, from roughly 774,000 to approximately 779,000 people [4], and payroll cost grew 7.3%, from $42,582M to $45,679M [5]. The revenue did not come from more people; it came from more value per person. But the variable that captures value per person is precisely the one management says it does not track.

Source: revenue and headcount from the FY2025 Form 10-K, Item 1 Business [6]; payroll cost from Note 16 [7].

The second proxy is utilization — how fully the people already on the roster are billed. It has run in a narrow band, 91% to 93% across FY2021–FY2025, and management has closed the door on it as a hidden lever. Asked directly on the Q4 FY2025 call whether internal AI would lift utilization structurally, the CEO said the company would "continue to move around in the low-90s. So we don't have a structural change in utilization due to AI." [8] A workforce billed in the low-90s cannot be billed much harder. So neither proxy is free to move: headcount grew less than one point in a seven-point revenue year, and utilization is against its ceiling. There is no third lever hiding behind the missing instrument. The only price series the company publishes is the operating margin itself — expanded in tens of basis points a year, a figure whose machinery belongs to the next act (Ten to Thirty Basis Points).

What the rivals are counting

The instrument Accenture says it lacks is one its named competitors have started to build and quantify. Cognizant's CEO described the change in the pricing basis in mechanical detail: "unlike in the past, where pricing was determined by the unit price, which is billing equivalent, the race now is about the number of units and how well you can deliver with a lower number of units for the same output… 40% of our software development cycle is assisted by AI. We have infused AI into our rate cards." [9] The unit of sale is migrating from an hour to an outcome, and AI-driven productivity is being priced into the rate card as it happens.

HCLTech went further and put a number on the deflation, then cut its own guidance for it. On its Q4 FY2026 call management said "the 3% to 5% deflation that I mentioned in the AI disrupted services… it would translate to 2% to 3% for our portfolio," added that "very little has really played out in already reported numbers," and guided FY27 revenue growth to "1% to 4% in constant currency" partly on that basis. [10] Set that against Accenture's framing of the same technology in the same twelve months: "We believe that AI will be a tailwind for us and our industry as it scales." [11] One management team names a percentage and lowers its plan; the other names a direction. Neither is provably wrong — but only one of them has published the number.

Whether that industry-wide deflation reaches Accenture's revenue depends on the form of its contracts, and here the company is more exposed than the rivals it out-scales. More than 60% of its work is fixed-price, a share that "continues to increase over 60% in FY25." [12] Under a fixed-price contract, a delivery-productivity gain accrues to Accenture as margin until the contract is renewed; under a rate-card contract it is passed through to the client at repricing. The 60%-fixed book is therefore a shelter for as long as the contracts run and a repricing event when they turn over — and Accenture discloses the share of fixed-price work but not its duration, so the moment the shelter lapses is not observable either. This is the live uncertainty the report will return to at the close: how much of the peer-quantified deflation reaches this particular book, and when.

Sources: Accenture FY2025 10-K [13] and Q2 FY2025 call [14]; Cognizant Q1 FY2026 call [15]; HCLTech Q4 FY2026 call [16]; TCS Q4 FY2026 call [17].

When the client becomes the competitor

The same pressure arrives through a second door, and this one Accenture and a peer describe independently. The FY2025 10-K's competition section lists, among the firms it competes with, "in-house IT departments of large corporations that use their own resources rather than engage an outside firm," naming global capability centres — client-owned offshore delivery units — as a competitor category. [18] Cognizant's 10-K frames the same category more sharply, as GCCs "which may provide a lower cost alternative to our services." [19] Infosys' FY2025 risk register books it as a quantifiable leak, warning that client GCC formation "may result in loss of addressable market share." [20] When the buyer can stand up its own delivery centre, the vendor's substitute is the customer itself.

Nothing in the contract form slows that substitution. The majority of Accenture's contracts, the 10-K notes, "are terminable by the client on short notice with little or no termination penalties, and some without notice." [21] The paperwork does not lock the client in. Whatever holds the relationship together, it is not a switching penalty written into the contract.

What holds the accounts

What does hold it is incumbency, and the evidence for it is specific and numeric rather than asserted. Accenture serves roughly 9,000 clients and reports it has "partnered with 195 of our top 200 clients for 10 or more years, and have 305 Diamond clients, our largest client relationships." [22] In FY2025 it booked a record 129 quarterly engagements above $100 million. [23] Decade-plus tenure across almost the entire top 200, and a rising count of nine-figure deals, is the strongest cited evidence that the largest reinvention work is won on the strength of the relationship, not the price of the hour.

The second protective asset is position in the technology ecosystem. Accenture reports it is "the number-one partner for all of our top 10 ecosystem partners," and that "60% of our revenue in fiscal 2025 was driven by work that we do with these partners, which grew 9%" — outpacing the company's own 7% growth. [24] That is a privileged place at the front of the queue when a hyperscaler or software vendor needs an implementation partner. It is also a concentration: well over half the revenue base is tied to the go-to-market choices of ten technology companies, so the asset and the exposure are the same fact read from two sides.

The distinction that matters is between the relationship and the revenue inside it. Incumbency and ecosystem rank protect the account — they keep Accenture in the room for the next transformation. They do not, on their own, protect the price of the work once that work can be delivered with fewer units, because the contract can be repriced at renewal even though the client never leaves. Stickiness of the relationship and durability of the unit price are two different questions, and the hard evidence answers only the first.

The soft spots in the outgrowth

Over five years the outgrowth is real and large. Accenture compounded revenue from $50.53B in FY2021 to $69.67B in FY2025, up 37.9% — well ahead of Cognizant's 14.0% over the same window, Capgemini's 23.7% and IBM Consulting's near-flat 0.8%.

Cumulative revenue growth: Accenture and Cognizant FY2021–FY2025 in USD; Capgemini CY2021–2025 in euros; IBM Consulting is the segment over the shorter FY2023–FY2025 window — the windows and currencies are not identical. Accenture's FY2025 revenue and 7% growth are as reported [25]; peer figures as reported in each company's annual filings.

Two things soften the share-gain story management tells on top of that record. The first is that the "market share" it claims is scored against a yardstick it defines itself. Management says it takes share "on a rolling four quarter basis against our basket of our closest global publicly traded competitors, which is how we calculate market share" [26] — a basket it elsewhere describes as "roughly two dozen of our closest global public competitors, which represents about a third of our addressable market." [27] That basket excludes the private Big Four and Deloitte, and excludes the client GCCs Accenture's own 10-K names as competitors, and no third-party share estimate appears anywhere in the corpus to test it. In FY2025 the company said it "took market share at more than five times our investable basket." [28] Over the same window Infosys reported "increased market share gains" [29], Cognizant said "we are gaining market share" [30], and TCS cited "market share gains" [31]. Four firms cannot all be taking share from one another; against a self-defined denominator, "gaining share" can be true for everyone and decisive for no one.

The second soft spot is that a material slice of the growth was bought rather than won. Management quantifies the inorganic contribution each year, and the arithmetic in FY2024 is the telling one: the company guided FY2024 to "1.5% to 2.5% growth in local currency… which assumes an inorganic contribution approaching 3%." [32] When the acquired contribution runs close to 3 points and the whole company grows about 2 points, the organic line went slightly backwards — the share Accenture reported taking that year was funded, not generated. FY2025 was healthier: of roughly 7 points of local-currency growth, "a bit more than 3%" was inorganic [33], leaving about 4 points organic. Management expects about 1.5% inorganic in FY2026 [34] and to enter FY2027 "slightly under at 2% of inorganic contribution." [35]

In FY2024 the acquired contribution (about 3 points) exceeded total local-currency growth (about 2 points), so organic growth was slightly negative. Source: Q3 FY2024 call [36], Q2 FY2025 call [37], Q3 FY2026 call [38]; FY2026 total is the midpoint of the guided 3–4% range.

A quieter tell sits underneath. The company's own count of its deepest relationships, the Diamond clients, climbed every year — 229, then 267, 300, and 310 at the end of FY2024 [39] — and then edged down to 305 in FY2025 [40], the first decline in the series, in a year revenue grew 7%. The metric then disappears: it is reported in none of the three FY2026 quarterly calls in the corpus. Full-year bookings slipped about 1%, and Q3 FY2026 bookings fell 3% in local currency at a book-to-bill of 1.0 [41] — the forward order book flat-to-down while the reported revenue line grew. The deepest-relationship indicator stopped compounding at exactly the point the company stopped publishing it.

What protects the relationship, and what does not

The contest over the unit price divides cleanly. On one side is what genuinely protects Accenture: decade-long incumbency across nearly the whole top 200, and rank as first-choice partner to the ten technology vendors that pull 60% of its revenue — both real, both numeric, both hardening on the disclosed evidence. On the other side is what does not: a market-share metric the company scores against a basket it draws itself, and a unit price it states it cannot measure, at the moment named rivals are quantifying that price falling 2–3% a year in the same book of work. Incumbency keeps Accenture in the account; it does not, by itself, keep the price of the hour inside the account from being renegotiated once the hour takes fewer people to deliver.

That leaves the one series the company actually publishes as the readout of price — the operating margin, which it has expanded in tens of basis points a year on the strength of exactly the fixed-price, productivity-retaining book described here. How that margin is held, what has been carried outside it to keep the promise intact, and how far the graded number has drifted from the statutory one, is the account that follows.