Sector-specific ratios

An analyst praises a software company for its impressive same-store sales growth. The sentence is nonsense - a software company has no stores - but it is the kind of nonsense spoken all the time in gentler forms: a hospital judged on the metric that belongs to a hotel, a bank praised for revenue growth while the number that actually matters quietly deteriorates, a new-age platform celebrated for a figure that says nothing about whether it makes money. Each time, a metric that is precise and powerful in its home sector has been carried into a sector it cannot describe, where it measures nothing and misleads confidently. illustrative

The generic ratios of this part - margins, returns, efficiency, leverage, valuation - apply, with care, almost everywhere. But every industry also has one or two hero metrics of its own: numbers that capture the essence of that business better than any generic ratio can, because they are built from the specific way it makes money. A generic ratio asks a general question, and a general question sometimes misses the specific thing that decides whether a business is winning. Return on capital tells you how hard a bank's capital works; it does not capture the single most important fact about a bank - the spread between what it earns on its assets and pays on its liabilities. Operating margin tells you a retailer's profitability; it does not separate growth that comes from genuinely selling more in existing stores from growth that comes merely from opening new ones. The hero metric encodes the actual economic engine of the business in a way the generic ratios cannot - and that same sharpness is exactly what makes it dangerous the moment it is misapplied. This module steps back from the sector modules earlier in the part to see these metrics as a set, and to learn the discipline of using each only where it belongs.

Each hero metric is a compressed statement of how its industry makes money. NIM captures that a bank's whole business is borrowing at one rate and lending at a higher one. ARPU captures that a telecom's value is the money each subscriber generates, not the raw count. Same-store sales captures a retailer's like-for-like growth, stripping out new-store openings. The combined ratio captures whether a general insurer's underwriting makes money before a rupee of investment income. ARPOB captures how much each occupied hospital bed earns. The way to hold these is not to memorise them but to see the engine each was built to measure - learn the engine, and the metric follows.

Most hero metrics travel in pairs. A single sector metric is usually necessary but not sufficient, because it captures one blade of a scissors. Load factor without yield cannot tell a profitable airline from one filling planes at fares below cost. Occupancy without ARPOB cannot tell a full hospital earning well from a full one earning little. GMV without contribution margin cannot tell a platform building a real business from one buying revenue at a loss. Subscriber growth without ARPU cannot tell valuable growth from price-cutting. So even inside its home sector, a hero metric is read with its partner - volume with value, scale with unit economics - and a management that quotes only the flattering half is telling you which half it would rather you not examine.

The hard rule: the metric stays home. Because a hero metric is built from the specific structure of one industry, it is meaningful only there. ARPU means nothing for a cement company; same-store sales nothing for a bank; the combined ratio nothing for a software firm. Carrying a metric across sectors is not a smaller version of a valid comparison - it is a category error that produces a number measuring nothing. So when you hear a sector metric used, the first check is whether it is being used in its home sector; the second, if so, is whether its partner metric is shown alongside it.

The whole point of this module is the across-sectors view, so here it is directly: the hero metric for four businesses, what it reveals at home, and the nonsense it becomes if carried elsewhere. The metric that reveals everything in one column measures nothing in the next.

This ties the whole part together. The generic ratios of the earlier modules are the common language you can speak, carefully, across most businesses; the hero metrics are the local dialects - unmatched at describing their own industry, gibberish anywhere else. A skilled reader uses both, and never confuses the two by carrying a dialect across a border. A ratio is a question, and a sector-specific ratio is a question only one sector can answer: ask it of the right business and it is the sharpest tool you have; ask it of the wrong one and it answers with confident nonsense.

A composite general insurer is presented two ways on the same day. Its own results release leads with "profit up 30%, investment income at a record, book value compounding nicely." A careful reader ignores all three and goes straight to the one number that decides whether the business works: the combined ratio. illustrative

Why skip the profit? Because a general insurer earns money two ways - by underwriting (collecting more in premiums than it pays in claims and costs) and by investing the float in the meantime - and the profit blends the two. A firm can report rising profit while its actual insurance business loses money, if a strong investment year covers an underwriting loss. The combined ratio strips that away: it is claims plus expenses as a percentage of premiums, and under 100% means the underwriting itself made money before a rupee of investment income. Read it here and it is 106% - the core insurance business is losing six paise on every rupee of premium, and the "30% profit growth" is entirely investment income masking an underwriting loss.

Now the discipline in the other direction. Suppose an enthusiast, impressed by the insurer, tried to compare it with a bank in the same financial-services basket by lining up their combined ratios. The comparison is meaningless - a bank has no combined ratio, because it does not underwrite insurance; its hero metric is net interest margin, an entirely different number built from an entirely different engine. So the live reading is two habits at once: reach past the generic profit number to the sector's hero metric to see whether the actual business works, and refuse to carry that hero metric one inch outside the sector it was built for. The insurer is judged on its combined ratio; the bank beside it on its NIM; and the two numbers never meet.

A hero metric is the sharpest single lens on a business, but a single lens, and leaning on it alone is its own trap. A bank's NIM can look healthy while the loan book quietly goes bad, because NIM measures the spread, not the credit quality behind it. A retailer's same-store sales can be strong while the company destroys value opening unprofitable new stores that the like-for-like number deliberately excludes. Read the hero metric alongside the generic ratios and the accounts, not in place of them.

Nor is a hero metric immune to being gamed or defined loosely. Same-store sales depends on which stores count as "same," and a company can flatter it by excluding weak stores from the base. ARPU can be lifted by reclassifying which customers count as subscribers. GMV can be inflated by counting cancelled or returned orders. Because these metrics are industry-specific and often not defined by accounting standards the way the statutory statements are, companies have latitude in how they compute them - so two firms' "same-store sales" or "ARPU" may not be built the same way, and the definition in the fine print matters as much as the number.

And the hero metric cannot tell you whether the sector itself is a good place to be. A telecom can post excellent ARPU trends in an industry being competed into the ground; an airline can improve its load factor and yield in a business that has destroyed capital for its owners for decades. The metric measures execution within the sector; it says nothing about whether the sector's structure allows anyone to earn a decent return over time. That larger question is one no ratio, generic or sector-specific, can answer - and it is where the final module of this part, and the later parts on competition and foresight, take over.

When management leads with the flattering half of a paired metric, a sharp analyst asks for the partner: "You've highlighted record subscriber additions. What happened to ARPU over the same period, and what was the blended ARPU on the new subscribers versus the base?" The analyst is refusing to judge volume without value.

A good answer supplies the partner and is honest about it:

"Reasonable to push on that. Gross additions were a record, but ARPU dipped about 3% because the new cohort came in on lower-value plans - that's deliberate, we're acquiring in smaller towns and expect to upsell over time. Blended ARPU on the base is stable. So yes, some of the subscriber growth is lower-value, and we'd rather you tracked ARPU alongside the additions, not the headline count alone."

The evasive version celebrates the volume metric and never mentions the value one:

"We added a record number of subscribers this quarter, cementing our market leadership. Momentum is exceptional across all circles, and we're very confident in our ability to monetise this expanding base going forward."

"Ability to monetise going forward" is a promise standing in for the number - ARPU - that would show whether monetisation is happening now. The follow-up that pins it down: "Give us ARPU alongside the subscriber growth, the contribution margin on the new cohort, and how you define a subscriber." If the room lets "record subscribers, exceptional momentum" stand without the ARPU trend, either the value per user is falling and the growth is being bought, or the subscriber definition is generous enough that the count flatters. A sector metric quoted in only its flattering half, with its partner and its definition left unspoken, is the setup this module teaches you to complete.

1. Carrying a hero metric across sectors. Same-store sales on a services firm, ARPU on a cement maker, the combined ratio on a software company - each is a category error producing a number that measures nothing. The metric is meaningful only in its home sector.

2. Reading half of a paired metric. Load factor without yield, occupancy without ARPOB, GMV without contribution margin, subscribers without ARPU - the volume half without the value half hides whether the growth is profitable. Read the pair, and treat the half a management chooses to quote as a signal of which half it would rather you not examine.

3. Leaning on the hero metric alone, or trusting its loose definition. A healthy NIM can sit over a book going bad; strong same-store sales can accompany value-destroying new stores. And because these metrics are rarely accounting-standard terms, companies have latitude in how they compute them - two firms' versions may differ, so the definition in the fine print matters.

4. Mistaking metric strength for sector attractiveness. Good execution on a hero metric says nothing about whether the industry lets anyone earn a decent return. Improving load factor in a value-destroying airline business is still value-destroying.