How we ship v1 in ten weeks.

We get asked, more than any other question, how a small studio can ship a real v1 — application, brand, distribution motion, in market — in ten weeks. The short answer is that we don’t really. The long answer is that we ship something that looks like v1 to a paying customer in ten weeks, and we are willing to be deeply uncomfortable for the entire stretch to make that happen.

This is the playbook. It is not a marketing artifact. It is the actual file we hand to a new founder in week zero, with a few names changed.

Why ten weeks (not eight, not twelve)

Ten weeks is the smallest amount of time in which we have repeatedly shipped and kept the team intact. Eight is heroic and the team breaks. Twelve is comfortable and the discipline goes; we start adding things we will end up removing.

Week one · pre-mortem

Before a single line of code, the studio runs a pre-mortem. We close our eyes and imagine it is week ten and the launch has gone badly. We then write down — in present tense, with concrete sentences — what went wrong. The pre-mortem is one page, no longer, signed by the founder and the studio lead.

Weeks two–four · the spine

The spine is the smallest possible end-to-end product that takes an input we expect and returns an output a customer would pay for. It is intentionally ugly. It is intentionally narrow. It is intentionally a single path. Brand, design polish, secondary flows, and admin tooling are all explicitly out of scope. The spine exists to disprove the wedge.

What we always do

1. 01
   One real customer in the build loop — Not a beta list. One person, by name, who has agreed to use the spine when it exists.
2. 02
   No design system yet — We work in raw shadcn / tokens-only mode until the spine works end-to-end. Brand sprint waits for week six.
3. 03
   Evaluations from day one — If the product calls a model, the eval pipeline exists before the second prompt does. Every change ships with a diff against the baseline.
4. 04
   A weekly demo, every Friday — Founder + studio lead + one outside observer. Fifteen minutes. The observer is the brake.

Weeks five–eight · in market

The spine ships to one customer by the end of week four. Weeks five through eight are spent using it with them — not building features for them, using it with them. We instrument every interaction. We sit in their workflow. Every Friday we cut whatever feature scored lowest on the customer’s actual usage that week, regardless of how attached the studio is to it.

How we measure

Every venture has exactly one north-star metric that the spine is supposed to move, agreed on in week one. Not three. Not “north-star plus health metrics.” One. Everything else is debugging.

# Studio standup, every Monday — six lines, no slides.
[ ] North-star metric, last week                   →  82 → 91 (+9)
[ ] What we shipped that we expected to move it    →  inline triage v2
[ ] What we shipped that didn't                    →  bulk import (delete)
[ ] What the customer actually did with it         →  daily, 4 sessions
[ ] What we are cutting this week                  →  admin search
[ ] What ships before Friday                       →  threshold alerts

Weeks nine–ten · earn it

The last two weeks are for the things only late weeks can earn: a real brand pass, a launch narrative, the second and third paying customer, the press list, the first hire. If you do them earlier they are speculative. If you do them now they are evidence-based.

When it slips (it slips)

Roughly one in three sprints slips by a week or more. The slip is almost always in weeks five through eight — the in-market stretch — when the product makes contact with reality and reality wins. When it slips, we do exactly three things, in order:

1. 01
   Cut something the studio is attached to. The slip is asking for a cut; do not negotiate with it.
2. 02
   Tell the founder. The slip is shared work; the founder is part of the recovery, not the audience for it.
3. 03
   Move the launch date publicly. The number of customers we lose by moving a launch is meaningfully smaller than the number we lose by missing it loudly.

Footnotes

1. “Spine” is a term we lifted from the IDEO product playbook, c. 2012, and have been quietly stretching to mean something a little different ever since. Sorry, Tim.

2. The “one real customer in the build loop” rule has exactly one exception, in case you are wondering, and it is when the founder is the customer. It happens twice a year and it ends fine.