What Is a Business? Value, Models & How Companies Make Money

Why this lesson exists

Almost everyone can describe what a company sells. Very few people can explain how it makes money — and those are different questions. Uber and Careem sell rides. Netflix sells shows. Those sentences are true and useless: they tell you nothing about whether either business survives a bad year, what happens to profit when they double in size, or why one of them can be copied in a month and the other cannot.

This lesson gives you the vocabulary and the method to answer the useful question. It installs one distinction, one framework and one taxonomy:

1. The distinction: value creation is not value capture.
2. The framework: five questions that describe any business model.
3. The taxonomy: the revenue model archetypes, and what each one does to your economics.

Everything else in this course — costs, elasticity, market structure, macro — is machinery for answering these questions more precisely.

---

1. A business is a machine for creating and capturing value

Strip away the branding and a business is a repeatable arrangement that does two things:

• Creates value: it makes someone better off than they were. A customer pays $45 for a meal delivered to their door because getting that meal themselves would have cost them more — in time, fuel, effort, or the simple impossibility of cooking it.
• Captures value: it keeps some of that gain for itself, as the gap between what the customer pays and what it costs the business to serve them.

Two boundaries define the space a business lives in:

Willingness to pay (the most the customer would pay)
        │
        │   ← consumer surplus (customer's share)
     PRICE
        │   ← captured value (the firm's share)
        │
     COST (what it takes the firm to deliver)

If price sits above cost, the business captures value. If willingness to pay sits above price, the customer gets a reason to come back. A durable business needs both gaps to be positive at the same time — and this is harder than it sounds, because competition pushes price toward cost and customer expectations push willingness to pay downward over time.

The failure mode you must learn to recognise: enormous value creation with no value capture. A free WhatsApp group that coordinates a neighbourhood's grocery orders creates real value — hours saved, better prices — and captures none of it. So did a decade of venture-funded delivery apps that gave customers $30 of convenience for $10 of revenue and $25 of cost. Creating value earns you gratitude. Capturing it earns you a business.

Ask, always: who is better off, by how much, and what share of that improvement do we keep?

---

2. The five-question business model framework

A business model is the answer to five questions. Answer all five with specifics and you can describe any company in under a page. Skip one and you have a slogan.

Q1 — Who is the customer?

Not a demographic. A person with a situation. "Working parents who need dinner solved on weeknights" is a customer. "Millennials" is not. Note also that the customer and the payer are sometimes different people — that is the whole design of advertising businesses, and it changes everything downstream.

Q2 — What value do they get?

Stated in the customer's terms and, ideally, in numbers. Not "convenience" but "saves 40 minutes on a weeknight". Not "premium experience" but "the wait is 4 minutes instead of 25". If you cannot quantify it, you will not be able to price it later.

Q3 — How is it delivered?

The activities, assets and partners required. This question quietly determines your cost structure, and cost structure is destiny. A business that delivers value through people has costs that grow with every customer. A business that delivers through software has costs that mostly do not.

Q4 — How do we get paid?

Who pays, when, how often, and how much. This is the revenue model (section 3). It is a design choice, not a fact of nature: the same value can be sold as a one-off transaction, a subscription, a commission, or given away and monetised through advertising.

Q5 — Why can't this be copied?

The question amateurs skip. If a competitor with money can replicate your model in six months, your margin has a short life. Real defences: network effects, switching costs, regulatory licences, proprietary data, brand, scale-driven cost advantage. Not defences: being first, working harder, "our team", a nice app.

---

3. Revenue model archetypes

The revenue model is how value gets captured. Each archetype carries characteristic economics.

Archetype	You are paid	Characteristic economics	Example
Transaction / unit sale	Once, per item sold	Revenue stops when selling stops; margin depends on COGS	Zara, a restaurant, a retailer
Subscription	Recurring, per period	Predictable revenue; economics live or die on retention	Netflix, Spotify, a gym
Marketplace / commission	A percentage of a third-party transaction	Asset-light, but you need liquidity on both sides	Uber, Airbnb, eBay
Advertising	By advertisers, for attention	User and payer are different; scale of attention is the product	Meta, YouTube, a free news site
Licensing / franchise	A fee for use of IP or brand	Very high margin, low control over execution	A hotel brand contract
Freemium	By the minority who upgrade	Free tier is a marketing cost; conversion rate is the whole model	Spotify, LinkedIn
Usage-based	Per unit consumed	Revenue scales with customer success; harder to forecast	Cloud hosting, utility metering

Two rules worth memorising:

• The same product can carry different revenue models, with wildly different results. Sell software once for $5,000 and you have a transaction business with lumpy revenue and a constant hunt for new buyers. Sell it for $200 a month and you have a subscription business whose value compounds — and whose survival depends on whether customers stay past month six.
• The revenue model determines which number you must obsess over. Transaction: gross margin per sale. Subscription: churn. Marketplace: take rate and liquidity. Advertising: engaged time and price per impression. Freemium: conversion rate. Get the wrong obsession and you will optimise a metric that does not pay you.

---

4. Worked example: Careem and Netflix through the five questions

Two companies, both successful, with almost opposite economics. This is the case you will study tomorrow, so here is the skeleton.

Careem

1. Who: someone in a Middle Eastern city who needs to get somewhere now, without owning a car or negotiating with a taxi.
2. What: a ride that arrives in about five minutes, at a price known before departure, paid without cash. Worth roughly the difference between that and standing on a road hoping.
3. How: a matching app, plus a supply of captains who own or lease the vehicles. Careem does not own the fleet; it owns the matching layer and the demand.
4. Money: a commission on each trip — a take rate on someone else's transaction.
5. Moat: local network effects (more captains means shorter waits, which brings more riders, which attracts more captains), city-by-city regulatory permissions, and habit.

The critical structural fact: Careem's costs are overwhelmingly variable. Serve one more ride and you pay one more captain. Doubling rides roughly doubles the payments out. The network is local — dominating one city gives you nothing in another, so the whole moat has to be rebuilt in every market.

Netflix

1. Who: a household anywhere with broadband and an evening to fill.
2. What: unlimited access to a large content library, no ads, no scheduling, one price.
3. How: license and produce content at very large fixed cost, then stream it globally over infrastructure whose marginal cost per stream is close to trivial.
4. Money: a monthly subscription, paid by the viewer, in advance.
5. Moat: content the customer cannot get elsewhere, scale that lets it spend more on content than smaller rivals can, and recommendation data.

The critical structural fact: Netflix's costs are overwhelmingly fixed. A series costs what it costs whether one household watches or eighty million do. The hundred-millionth subscriber is nearly pure profit — which is precisely why the business is a race for scale, and why a stall in subscriber growth is treated as an emergency.

The lesson. Careem's model is hard to scale but hard to break in the city it owns. Netflix's model scales magnificently and is permanently exposed to anyone with a bigger content budget. Neither is better. Each is a coherent set of trade-offs — and reading those trade-offs is the skill this course is teaching you.

---

5. How to read a business from the outside

You rarely get the internal numbers. You can still get most of the way there with observable evidence:

• Watch the transaction. Who hands over money, how often, and for what unit? That is Q4, answered in five minutes at any shop.
• Count the people. How many staff are involved in serving one customer? That estimates how variable the cost structure is.
• Look for the asset. Do they own the vehicles, the kitchen, the inventory, the building — or do they broker someone else's? Asset-light and asset-heavy businesses fail in different ways.
• Test the switch. How hard would it be for their customer to use the competitor next week? The answer is the moat, measured honestly.
• Find the payer. If the user is not paying, someone else is, and the business is built for them.

Tomorrow you take Careem and Netflix apart properly. On Day 3 you go outside and do this to a business you can actually watch.

---

Check your understanding

Answer these in writing before Day 2. No answers are provided — the point is to produce your own and test them against tomorrow's case.

1. Give an example of a service that creates substantial value and captures almost none of it. Explain precisely where the captured value leaks away.
2. A gym charges $300 a month. Identify its revenue model archetype, and name the single metric its survival depends on most. Why that one?
3. Careem and Netflix both have millions of customers. Why does adding one million more customers affect their profit so differently?
4. Answer Q5 for a coffee shop you know. If your honest answer is "nothing much protects it", what would you have to change about the business to create a real defence?
5. A company sells a training course for $4,000 once. Redesign it as a subscription. What breaks, what improves, and what new number would you have to watch weekly?
6. The customer and the payer are different in advertising businesses. Name one consequence of that split that shows up in the product itself.

---

AI toolkit: using AI on business models, responsibly

AI is genuinely good at the structural part of this work and genuinely unreliable at the factual part. Use it accordingly.

Good uses this week

• Ask it to apply the five questions to a company you have described, then argue with its answers. Its Q5 answers in particular are usually generous — a good place to practise scepticism.
• Ask for five alternative revenue models for a given product, then evaluate each yourself on cost structure and the metric it would force you to obsess over.
• Paste your own business model description and ask: "Which of these five questions have I answered vaguely?" It is a reliable vagueness detector.

Where it will mislead you

• Take rates, market sizes, subscriber counts, prices and dates. Models produce confident, specific, wrong numbers. Every figure that enters your work needs a source you have seen with your own eyes.
• Moats. AI will tell you a company has "strong brand loyalty and network effects" about almost anything. Demand the mechanism: who is locked in, by what, and what it would cost them to leave.

The rule for this course: AI may shape structure and generate alternatives. It may not supply facts or final judgements. When you use it in a deliverable, state in one line what you asked, what you kept, and what you threw away and why.