One toolkit, many businesses

This series built one toolkit — margins, returns, working-capital days, leverage, valuation multiples — and tested it on one kind of business: a packaged-food maker (the fictional Desi Bites and the real Britannia). The banks module then showed that a lender breaks half of that toolkit.

Banks aren’t the only exception. Every sector has a few numbers that explain more about its economics than any generic ratio, because they measure the thing the business actually sells. A sector lens is that short list of metrics — often called sector KPIs (key performance indicators) — used alongside the standard ratios, not instead of them.

The generic ratios still work everywhere; the lens tells you what they mean. A 20% EBITDA margin in a software company, a cement company and a telecom company describe three very different businesses.

The formula behind every lens

Almost every sector lens is the same move: split revenue (or profit) into how many units and how much per unit, using the unit the business really sells.

Revenue  =  Units sold  ×  Revenue per unit

  IT services     billable people × utilisation × billing rate
  FMCG            volume (tonnes, packs) × price and mix
  Cement, metals  tonnes sold × realisation per tonne
  Telecom         subscribers × ARPU (average revenue per user)
  Lenders         loan book × spread (yield − cost of funds)
  Life insurers   new business (APE, annualised premium equivalent)
                  × margin on it (VNB margin, value of new business ÷ APE)

The generic ratios sit on top of that split. EBITDA margin (earnings before interest, tax, depreciation and amortisation, as a share of revenue) tells you what’s left per rupee of sales. The sector lens tells you why — more units, a better price, or a cost per unit that fell.

Worked example: the same two ratios, four businesses

Figures from each company’s own results release for the year to 31 March 2026 (FY26, released April–May 2026) and, for Britannia, the FY25 filing this blog already uses. Sources: Britannia, UltraTech Cement, Dr. Reddy’s, Bharti Airtel. Historical, for illustration only — and note the periods differ, which is fine for seeing structure but not for ranking.

Company (sector) Period EBITDA margin EBIT margin Capex as % of revenue
Britannia (FMCG) FY25 17.8% 16% 2.1%
UltraTech (cement) FY26 20.1% — 11%
Dr. Reddy’s (pharma) FY26 22.8% — 6.8%
Bharti Airtel (telecom) FY26 (capex: Q4 only) 57.5% 32.3% 29%

EBIT is earnings before interest and tax — EBITDA after depreciation and amortisation. UltraTech’s release gives PBIDT (profit before interest, depreciation and tax, its name for EBITDA) but not depreciation, and Dr. Reddy’s release doesn’t print an EBIT line, so those cells are left blank rather than estimated. Airtel’s capex figure is for the March 2026 quarter only, against that quarter’s revenue.

Read the table one column at a time.

  1. EBITDA margin. Airtel’s is about 3.2 times Britannia’s. On this number alone the telecom company looks like the far better business.
  2. EBIT margin. After depreciation — the cost of the towers, fibre and spectrum already bought — the gap narrows to about 2 times. A network business earns its EBITDA by spending heavily up front, and depreciation is where that spending comes back.
  3. Capex. Airtel spent about 29% of the quarter’s revenue on new capital assets; UltraTech 11% of the year’s; Britannia 2.1% (see capex intensity). A high EBITDA margin that has to fund a network every year is a different thing from a lower one that mostly turns into free cash.

None of this says which company is better. It says the same ratio is measuring different things, so it can’t be compared across the table.

Returns: a bank and a biscuit maker

The same problem hits returns. Britannia’s ROA (return on assets) for FY25 was 24.3%. HDFC Bank’s for the same year was 1.79% — Britannia’s is about 14 times as high. That isn’t a verdict on either; it’s what the banks module explained. A bank’s assets are loans funded mostly with other people’s money, so it earns a thin return on a huge balance sheet and multiplies it with leverage (about 8x assets to equity), landing at ROE (return on equity) of 14.4%. A biscuit maker’s assets are ovens and inventory it owns outright. ROA is comparable within banks, or within manufacturers — not between them.

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

In cricket, you don’t judge a fast bowler by his batting average, or an opening batter by how many wickets she takes. Both numbers are real. They’re just measuring the wrong job.

Companies are the same. A phone company’s job is to build a network and fill it with customers. A biscuit company’s job is to sell lots of packets at a decent price. A bank’s job is to lend money that mostly isn’t its own. You need a different scorecard for each — and then you can still compare two bowlers, or two banks, with each other.

The lenses in this module

Each sector post that follows takes one real company’s FY26 filing and works through the metrics that sector lives by. The companies are there to make the numbers concrete, not as views on the companies.

Sector What the business really sells Metrics the lens adds Anchor filing
IT services People’s time, billed in dollars Constant-currency growth, utilisation, attrition, deal wins Infosys, FY26
FMCG (fast-moving consumer goods) Packs on shelves Volume growth vs price growth Marico, FY26
Cement Tonnes from a fixed plant EBITDA per tonne, capacity utilisation UltraTech, FY26
Pharma Molecules, patents and approvals R&D (research and development) intensity, US generics vs India mix, US FDA (Food and Drug Administration) inspections Dr. Reddy’s, FY26
Life insurance Long contracts priced today, paid out over decades Embedded value, VNB margin, persistency HDFC Life, FY26
NBFCs (non-banking financial companies) Credit, funded without current and savings accounts Cost of borrowing, spread, asset-liability matching Bajaj Finance, FY26
Telecom Network capacity, rented monthly ARPU, EBITDA after leases, capex Bharti Airtel, FY26

Two sectors already have a module: banks (above), and the packaged-food business this series has used from the start.

Common mistakes

  • Screening the whole market on one ratio. A filter like “EBITDA margin above 30%” pulls in capital-heavy businesses such as telecoms, where much of that margin goes back into capex. A filter like “ROA above 10%” won’t show you a bank. Screen within a sector, or use a ratio that already accounts for capital, such as ROCE (return on capital employed) — and even then, compare within the sector.
  • Comparing with “the market average” instead of the right peers. A sector average hides very different business models. Two pharma companies can have opposite exposure to US generic pricing; two IT companies can earn in different currencies. The peer set is part of the analysis (relative valuation made the same point about multiples).
  • Treating company-defined metrics as standard. “PBIDT”, “operating PBIDT per tonne”, “underlying volume growth”, “before one-time actions” — each company chooses its own definitions, and many of these numbers sit outside the audited statements. Before comparing two companies’ versions, read the footnote that defines each one.
  • Letting the lens replace the basics. A sector KPI explains a company’s economics; it doesn’t replace the balance sheet, the cash-flow statement or the red-flag checks from earlier in this series. A cement company with a great EBITDA per tonne can still be over-borrowed.

Takeaway: The same ratio measures different things in different businesses, so it can only be compared within a sector. Each sector has a few metrics — usually units sold and money made per unit — that explain what the generic ratios can’t; learn those, then read the standard ratios through them.