The routes and rules
6 routes, each a direction of travel with its next steps, and the 8 rules that pick one from the signals, applied in order, first match wins, the last one a default. The record shows which rule fired, so you can disagree with the rule rather than with a verdict.
The rules
| Rule | If | Then | Because |
|---|---|---|---|
| R1 | entry is blank | Start the loop | No startup yet: the loop is the route. |
| R2 | cost is capital | Go old school | Capital before a first user. |
| R3 | raise is must | Go old school | No first user without raising. |
| R4 | entry is shipping | Give it away, take it away | Already shipping: the miss test is next. |
| R5 | entry is many | Zoom in | Several ideas: pick the slice. |
| R6 | scope is platform | Zoom in | A platform is several startups. |
| R7 | builder is component or framework and exists is nothing or demo | Ship the component | A component builder with no product yet ships the component. |
| R8 | otherwise | Ship it this week | Everything else can ship this week. |
Deterministic on purpose: the same answers give the same route on the page, in a model and in the test. What a model adds is the conversation that gets to honest answers, and the record in your words.
The routes
Start the loop start-the-loop
When: No idea yet.
There is no startup yet, and that is fine. The route is the loop itself: make the smallest thing somebody you know could use, from what you already build and know, hand it to them, and see what they do with it. The startup is found by shipping, not by thinking about shipping.
- Write down what you naturally build and three people you could make something for.
- Pick the one thing you could put in their hands in two weeks.
- Come back to the finder with that as the thing.
Go old school old-school
When: The standing cost is capital, or there is no first user without raising.
This one cannot reach a first user without capital: hardware, a licence, a regulated market, people on day one. There is no shame in it, and no shortcut either. The route is to raise, which means the plan has to be the product for a while: the evidence a stranger can check, the smallest capital that reaches the first user, and the terms you can wait for.
- Find the smallest thing that can be shipped before the capital: a study, a waiting list, a paid pilot, a component.
- Write the plan as a vault somebody else could check: facts apart from assumptions.
- Be the one the investors call: ship that smaller thing first.
Give it away, take it away next-step
When: Already shipping to strangers.
Already shipping. The next step is the test almost nobody runs on purpose: hand it to five people for a short window, take it away, and ask whether they miss it. If they ask for it back, price it. If they shrug, you have learnt the most useful thing a startup can learn, early.
- Name the five, the window and the take-away date.
- Write the signal that counts as missing it before the window opens.
- If they miss it: a price, in the shape the money question chose.
Zoom in zoom-in
When: Several ideas, or a scope of platform.
Too wide to ship. Several ideas, or one idea that is a platform. The route is to pick the slice that one person can use on day one and ship that, with the rest written down as what waits. A platform is what a product becomes after it has been missed.
- List the slices; for each, who uses it and what they do on day one.
- Pick the one with the shortest path to a stranger's hands.
- Put the rest in the not-doing list, dated.
Ship the component component-first
When: A component or framework builder whose product does not exist yet.
You build components, tools or infrastructure, and the product is not there yet. That is a normal path; the author of this site arrived at his product startups a few components later. The route is to ship the component as the thing: open source, usable by another builder this week, with a read key or a package name as the whole distribution.
- Name the component another builder would use on its own.
- Publish it where they already look: a package, a repository, a vault.
- Watch who uses it; the product is in what they ask for next.
Ship it this week ship-now
When: Everything else.
It can be done now. Whatever the blocker is called, the route is the same: the smallest version a stranger can use, in their hands by the date on the record, with the five people and the take-away date already written down. It is not ready yet is the problem, not the reason.
- Ship the smallest version to the five by the date on the record.
- Run the miss test as written.
- Come back with what happened; the record gets a second version.
Old school is a route, not a failure: some startups cannot reach a first user without capital, and the honest thing is to say so and plan for it. Ship the component is the author's own path, three or four startups and a number of components later.