Choose efficiency rather than growth by default
⏱ Untimed topicFried describes 37signals as focused on efficiency, small teams and customers it knows rather than maximizing growth. He links that choice to fewer offerings and less organizational complexity. This is his business model, not evidence that larger teams are always worse.
Keep the product surface simple
⏱ Untimed topicHe describes one code base and a shared product rather than many customized enterprise versions. Different offerings, service expectations and sales processes can create costs beyond the feature itself. The prices and staffing examples in the interview are historical claims, not current product information.
Ask whether the work is worth doing again
⏱ Untimed topicFried's practical success test is whether a project was worthwhile and enjoyable enough to repeat. He pairs that with the need for the business as a whole to make more than it spends. This does not remove financial responsibility or imply that every individual activity must have a separately measurable return.
Use an appetite rather than an open-ended estimate
⏱ Untimed topicIn Shape Up, he describes six weeks as the maximum budget for a feature, not a prediction of how long an unlimited scope will take. Teams seek a simpler effective version within that budget. Smaller work can have a shorter appetite rather than automatically receiving six weeks.
Shape work and let the small team solve it
⏱ Untimed topicFried describes shaping an idea ahead of time, then giving a designer and programmer room to decide the work needed. Scope and trade-offs can change during the cycle. His account includes contextual exceptions near completion, rather than an absolute rule that every unfinished item is discarded.
Be careful with distant delivery promises
⏱ Untimed topicHe says promises such as by the end of the year are easy to make and often regretted. He recommends trying a new working method on low-criticality work before changing everything. His no-promises view is advice to evaluate, not permission to abandon existing commitments.
Use cool-downs for other kinds of work
⏱ Untimed topicHe describes a typical two-week cool-down after six-week cycles for bugs, finishing touches and internally chosen work while the next cycle is shaped. It is not two weeks without work. He also says data can be an input to judgment, while decisions still draw on experience and other factors.
