The same job, bent by the market it runs in. Seven sectors, each a full product treatment rather than a list of industry facts: who actually buys, what the cycle does to the loop, what the product must contain that no user ever asked for, where a model is allowed to decide, and what that does to the roadmap.
Read AI Product Shapes first. The shape of a product usually determines more than the market it sells into. A fintech marketplace behaves like a marketplace that happens to move money, and borrowing the fintech playbook for it will mislead you about liquidity, which is the thing that will actually kill it.
| # | Lesson | The question it answers |
|---|---|---|
| 01 | Enterprise B2B SaaS | Why does the buyer want things the user will never see, and what does that cost? |
| 02 | Fintech | What changes when being wrong moves money and a regulator has an opinion about how? |
| 03 | Ecommerce and D2C | Why is this a margin business wearing a funnel costume? |
| 04 | Marketplaces | How do you build for two customers whose interests oppose each other? |
| 05 | Devtools and Infrastructure | What happens when your users can read the source and will tell you why you are wrong? |
| 06 | Healthtech and Regulated Care | How do you build around a clinician when the buyer, the payer and the patient are three people? |
| 07 | Vertical SaaS | How do you win a small market completely, and when does services gravity eat you? |
The comparison matrix#
The five structural forces, across all seven. This table is why the sector block exists in one place: every other discussion states its deltas against these rows and never restates them.
| Who signs | Cycle | Regulation | Data constraint | Margin shape | Abuse surface | |
|---|---|---|---|---|---|---|
| Enterprise B2B SaaS | A committee, with procurement and security | Two to twelve months | Contractual rather than statutory, plus certifications | Tenancy, residency, retention, audit | High gross margin, very high cost of sale | Low, but credential and permission misuse matters |
| Fintech | Varies, but risk and compliance hold a veto | Weeks to a year | Statutory, with licensing and supervision | Residency, retention, lineage, right to explanation | Spread or fee based, thin and volume driven | Very high: fraud, laundering, account takeover |
| Ecommerce and D2C | One person, in minutes | Minutes | Consumer law, tax, payments | Mostly volume and freshness | Thin contribution margin per order | Payment fraud, returns abuse, review manipulation |
| Marketplaces | Both sides separately | Minutes to weeks | Varies by what is traded; often employment or tax | Identity and trust signals on both sides | Take rate on gross volume | Very high: disintermediation, fake supply, scams |
| Devtools and Infrastructure | A developer first, finance later | Days to adopt, months to pay | Light, until you hold customer data | Logs, secrets, telemetry consent | High margin, except where you resell compute | Quota abuse, crypto mining, supply chain |
| Healthtech and Regulated Care | An institution, rarely the user | Six to twenty four months | Heavy and clinical, with device regimes | Strict privacy, consent, residency, retention | Varies wildly by reimbursement | Identity, prescription and claims fraud |
| Vertical SaaS | An owner or operator, often non technical | Weeks to months | Whatever the industry carries | Industry specific records and retention | High margin, dragged down by services | Low, but the data is often the business of record |
Treat these as orientation, not benchmarks. They are the typical shapes seen as of 12 October 2026, and they are deliberately ranges. Real figures, with sources and dates, live in Benchmarks and Their Limits.
How to read a sector you have never worked in#
Six questions, in this order. The order matters, because each answer constrains the next.
- Who signs, and who uses? If they are different people, almost everything else follows from that single fact.
- What is the cycle, end to end? This sets your loop length, which sets how fast you can learn.
- What is forbidden? Regulation and contract define the envelope before you design anything.
- What does the data layer permit? Residency, retention and consent rules kill features early and cheaply if you know them, expensively if you do not.
- Where does the margin come from? A thin margin business cannot afford the same product decisions as a high margin one, regardless of how good the feature is.
- How will it be abused? Every sector has a dominant abuse vector, and the first version of your product will meet it.