Every industry speaks its own dialect of metrics: ARPU here, RevPAR there, combined ratio somewhere else. You can't memorize them all, and that isn't the skill anyway. What transfers is a small taxonomy of KPI families. The family, unlike the local name, recurs in every industry you'll ever meet, including ones nobody has written a playbook for yet:
| KPI family | Generic definition | Two examples (local name) |
|---|---|---|
| Volume × Price/Rate | units or transactions × the price or rate charged per unit | subscribers × ARPU (telecom); passengers × yield (airlines) |
| Utilization / Efficiency | the share of available capacity actually used or turned | occupancy % (hotels, multiplexes); inventory turns (retail) |
| Unit economics / take rate | contribution or margin earned per repeatable unit | contribution margin ladder (e-commerce); take rate (marketplaces) |
| Quality / Loss / Risk | the rate at which the business fails its own promise | combined ratio (insurance); NPA/GNPA ratio (lending) |
| Growth & retention | how much of last period's base or revenue survives, and how fast new base is added | net revenue retention (SaaS); same-store sales / comps (retail) |
| Structural / regulatory ratios | a ratio a regulator or lender imposes as a condition of operating | capital adequacy ratio (banks); net debt/EBITDA covenant (leveraged industries) |
Advertisement
How do you find the right KPIs for an unfamiliar industry?
So when you meet an unfamiliar industry, don't start by hunting for its specific metric names. First ask which of these six families does this business's economics actually run on? Only then go and find that family's local name and healthy range. The worked example on KPI triage in Indian multiplex exhibition shows the move on an industry picked because most learners have never studied its metrics before.
{
"skill": "Compressing the blank-page cost of a first-pass KPI panel and regulatory overlay for an industry you have never studied",
"use": "Have a chat assistant draft a candidate KPI panel (which of the six families the business likely runs on, and the metric's probable local name) and name the likely regulator(s) and what they control, as a research map — then trace every claim to a primary source before it enters your one-page note.",
"tools": ["Chat assistants — Claude, ChatGPT, Gemini", "Research/agentic tools — a browsing-enabled assistant for regulator sites and recent filings", "Document/PDF AI — run over an annual report or 10-K you already hold"],
"prompt": "I'm reading the <INDUSTRY> industry cold. Propose the 5-8 KPIs practitioners actually track, tagged by which of these six families each belongs to: volume×price, utilization, unit economics, quality/loss, growth/retention, structural/regulatory. Name the primary regulator(s) and what they control. For every claim, tell me exactly where I should go verify it — do not state a 'typical range' as if it were sourced.",
"verify": "Treat this as a research map, never a source. Run the Primary-Source Guardrail (Trace, Match, Recompute, Date-check, Cite — AI0.01) on every number and regulatory claim before it reaches your note: trace each KPI's typical range to the sector's own filings or the nearest finance-galaxy playbook (M5.01-M5.10), and confirm the regulator and rule are current — regulatory facts date fast, and an AI-drafted 'typical range' is never itself the citation.",
"diy": "You must already be able to run all eight steps by hand on a cold industry with nothing but public filings and a search engine — that is this module's actual gate. AI only compresses the first hour of desk research; it never replaces the read.",
"market": "IN"
}