← Back to Blog

Six-Week Cycles: The Deadline You Agree to Before the Design

The Compass Team

November 22, 2026

Six weeks is long enough to finish something meaningful and short enough to feel the deadline from the beginning. That line sits in the glossary of Shape Up, the book behind Basecamp's six week cycles, and the whole method hangs off it.

The odd part is what happens when the six weeks run out. In most companies, a project that isn't finished gets more time. In Shape Up it gets cancelled by default.

The book defines the move as a circuit breaker: cancel projects that don't ship in one cycle by default instead of extending them by default.

An extended deadline is a forecast you revise. A deadline with a circuit breaker behind it is a decision.

What makes the six weeks mean anything sits upstream of the cycle, in one number. Appetite.

An appetite answers a different question than an estimate does. Singer draws the line in his chapter on boundaries. "Estimates start with a design and end with a number. Appetites start with a number and end with a design. We use the appetite as a creative constraint on the design process."

That inversion is the mechanic. Asking how long something will take hands the design a blank check. Asking what the problem is worth gives the design a budget, and the design changes shape to fit it.

Basecamp sets appetite in two sizes. A small batch is something one designer and one or two programmers can build in one or two weeks. A big batch takes the same team a full six weeks.

Anything that can't be narrowed into a six week frame gets cut down until it can.

The case study in that chapter is a customer asking for a calendar. The team didn't ask what the calendar should look like. They asked when she wanted it.

She described driving to the office to check a wall calendar because a client called while she was working from home, and by the time she arrived there was no free room anyway.

The request was do everything a calendar does. The need was show me the free spaces. Basecamp shipped a grid of dots.

An appetite also settles fights about quality. Singer's version: without a time limit there is always a better version, and when you are hungry and in a hurry, a hot dog is perfect.

Six weeks is the hurry. The hot dog is the version you can finish.

Before the cycle starts, the work has to be shaped into a pitch, and the pitch carries five parts: the problem, the appetite, a rough solution that fits inside it, the rabbit holes, and the no-gos.

Founders skip the last two. Rabbit holes are the parts too open-ended to bet on. No-gos are what the solution deliberately won't do.

Writing down what you refuse to build is the same act as writing down what you decided. Both are records, and both are what someone can check you against nine months later.

Here is the honest part. Shape Up doesn't always stick. Scale X ran it for two years and wrote up why the team switched back, and two of their reasons are worth reading because both are record failures rather than method failures.

First, appetite got read as a target duration. When they set a three week appetite, the work expanded to fill three weeks.

That is the estimate sneaking back in wearing the other word. An appetite is a ceiling, and the moment it is treated as a schedule it stops constraining anything.

Second, they wouldn't delete code at the end of a cycle when a project wasn't shippable. The deadline lost its potency. A circuit breaker only works when it is a decision someone wrote down, and they had quietly recorded theirs as a preference.

What they kept is also worth copying. Their pitches carried a falsifiable hypothesis about expected usage or impact shortly after launch. That forced the conversation onto value instead of implementation detail.

A bet with a predicted outcome is a record. A bet without one is a hope with a start date.

A solo founder doesn't need a betting table or a two week cool-down. The transferable part is four lines written before the work starts: what this is worth in time, the smallest cut that would still ship, the hypothesis being tested, and what happens if it isn't done by the end of the cycle.

That last line decides whether the other three are real.

The number that actually runs the cycle is the one you wrote before you knew how you would build it.

Share this article