What you’re really analysing

An Indian IT services company sells its people’s time and skills — writing, testing and running software, and increasingly running whole processes — to clients who are mostly abroad. That one sentence explains the sector lens:

  • It earns in dollars, euros and pounds, but reports in rupees, so a weaker rupee lifts reported growth without a single extra hour of work.
  • Its capacity is people, so how many of them are billed (utilisation), how many leave (attrition) and how many it hires drive both revenue and margin.
  • Its future revenue is signed in advance, in multi-year contracts, so deal wins are the closest thing to an order book.

The generic ratios still apply — Infosys’s operating margin and debtor days mean what they always meant. The lens is about reading the top line correctly.

The formulas

Growth in rupees   ≈  (1 + growth in US$) × (1 + change in ₹ per US$) − 1

Growth in US$       =  Constant-currency growth  +  cross-currency effect

Constant-currency (CC) growth
   = this year's revenue in each billing currency, converted at LAST year's
     exchange rates, compared with last year's reported revenue

Utilisation            =  effort billed to clients / total available effort
Attrition (LTM)        =  employees who left in the last twelve months / average headcount
Revenue per employee   =  revenue / average headcount
TCV (total contract value) = everything a signed contract is expected to bill,
                             over its whole life

Constant currency is the company’s number for “how much more work did we sell”, with exchange rates held still. Utilisation and attrition definitions vary a little between companies (with or without trainees; voluntary or all exits), so always read the footnote.

Worked example: Infosys, FY26

All figures from Infosys’s own IFRS press release, US-dollar release and Fact Sheet for the year ended 31 March 2026, released 23 April 2026 and filed with the US SEC (Securities and Exchange Commission). IFRS is International Financial Reporting Standards, the basis Infosys uses for these releases. Historical, for illustration only.

Step 1: one year, three growth rates

FY26 FY25 Growth
Revenue, US$ million 20,158 19,277 4.6%
Revenue, ₹ crore 1,78,650 1,62,990 9.6%
Implied ₹ per US$ (₹ revenue ÷ US$ revenue) 88.62 84.55 4.8%
Constant-currency growth (as reported)     3.1%

The same year’s business grew 3.1%, 4.6% or 9.6%, depending on which number you read.

  1. Rupees vs dollars. Each dollar of revenue was worth about 4.8% more rupees in FY26 than in FY25 (an implied average of ₹88.62 against ₹84.55). The check: 1.046 × 1.048 = 1.096, against rupee growth of 1.096.
  2. Dollars vs constant currency. The remaining 1.5 percentage points came from revenue billed in other currencies being worth more dollars. The Fact Sheet shows where: in the March 2026 quarter, Europe grew 11.4% reported but 4.1% in constant currency.
  3. What’s left. About 3.1% is the growth that came from selling more work (or better-priced work). That is the number the company itself leads with, and guides on.

So of the 9.6% rupee growth, roughly 6.5 points was currency. That flows into rupee profit too — which is why an IT company’s rupee EPS (earnings per share, see the EPS post) can look strong in a year of modest real growth.

Step 2: the people engine

From the Fact Sheet, quarter ended 31 March, IT services:

Mar 2026 Mar 2025
Total employees 3,28,594 3,23,578
Utilisation, excluding trainees 83% 84.9%
Utilisation, including trainees 79.7% 81.9%
Voluntary attrition (last twelve months) 12.6% 14.1%
Effort delivered onsite (at client locations) 22.8% 23.6%
  1. Headcount rose by 5,016 (1.6%), less than constant-currency revenue’s 3.1%. Yet utilisation fell (next point), so the extra revenue per person didn’t come from billing more of each person’s hours — it came from pricing, mix or getting more done per hour on fixed-price work. The filing doesn’t split those.
  2. Utilisation fell 1.9 points (excluding trainees). Fewer of the available hours were billed. That can mean softer demand, or capacity built ahead of expected work; the number alone doesn’t say which.
  3. Attrition fell 1.5 points. Fewer people leaving means lower hiring and training costs and less wage pressure — and, in a slow market, it can also simply mean fewer outside offers.
  4. Revenue per employee: US$20,158 million over an average of about 3,26,086 people is roughly US$61,800 each (about ₹54.8 lakh). It’s crude, but tracked over time it shows whether each person is producing more.

Step 3: the order book

  1. Large deal wins had a TCV of US$14.9 billion in FY26, 55% of it “net new” (the rest being renewals of existing work). That’s about 0.74 times the year’s revenue — but it will be billed over several years, and part of it replaces revenue that was already there.
  2. Concentration and collection. The top ten clients were 20.2% of revenue; days sales outstanding (DSO) was 67 days, against 69 a year earlier.
  3. Margin. Reported operating margin was 20.3% against 21.1% in FY25. The company also shows 21% “adjusted”, excluding a one-off ₹1,289 crore charge from India’s new Labour Codes — the kind of adjustment the exceptional items post says to read, not just accept.

None of this is an opinion on Infosys. It’s one year’s filing, read through the sector’s own metrics.

🧒 Explain it like I'm 10 (optional — skip if this is already clear)

You sell lemonade to a friend in America for $1 a cup. Last summer $1 got you ₹84 at home. This summer it gets you ₹88.

You sold exactly the same number of cups — but you “made” more rupees. Did your lemonade stand get better? No. The dollar just bought more rupees.

Constant currency is asking: “If the dollar were still worth ₹84, how much more did I really sell?” That’s the part you actually earned by making more lemonade.

Common mistakes

  • Reading rupee growth as business growth. In FY26 about 6.5 of Infosys’s 9.6 points of rupee growth came from currency. When the rupee strengthens, the same effect runs in reverse. Start from constant currency.
  • Treating deal wins as next year’s revenue. TCV is spread over the contract’s life, may include renewals, and can be cut if the client’s plans change. Compare it with revenue over several years, and pay attention to the net-new share.
  • Reading utilisation and attrition as “higher is better” or “lower is better”. Very high utilisation leaves no spare people for new projects; low attrition can reflect a weak job market rather than happy staff. Both are signals to explain, not scores.
  • Comparing metrics with different definitions. Utilisation with and without trainees differs by several points; some companies report voluntary attrition, others total; some annualise a quarter, others use the last twelve months. Line up the definitions before lining up the numbers.

Takeaway: For an IT services company, start with constant-currency growth — rupee growth mixes the business with the exchange rate. Then read utilisation, attrition and deal wins as signals about demand and capacity, not as scores, and check each company’s definitions before comparing.