The 10-week sprint: pick one number, then audit your week against it

Darius Monsef has been through Y Combinator twice. His summary of what the programme actually does is short: ninety days of high pressure, focus on the most important thing, set a clear and measurable goal, and every week focus on that, with someone honest holding you to it.

He built that structure into a method he uses with Founders, and later into a product. It runs on ten weeks.

Week zero: choose the number

The sprint opens with one question. Which single number, if it changed, would mean this company is meaningfully further along than it is today.

One number. Not a dashboard. The point is to force a choice about what the business actually needs, before the calendar fills with work that feels productive.

That number then gets chunked into milestones, because ten weeks is already long enough that people struggle to reason about it as a single block.

Every week: highest impact only

Each week starts by identifying the work most likely to move that number. Everything else waits.

End of week: the audit

This is the part most goal-setting skips. At the end of the week, activity gets checked against results.

Darius gives a worked example. A Founder says they will talk to 15 Founders this week. Fine. Three weeks later, they have talked to 45 Founders and the sales numbers have not moved. The activity was real and the output was zero, and without a weekly audit that pattern can run for months.

His line on it: busy feels like progress, and it really is not.

Set the number before you need it

The mechanism only works if the target is agreed in advance. Ten more weeks, 100 more users, and if we cannot get there then it is not working.

Set after the fact, the number bends. Whatever happened becomes the thing you were aiming at, and the narrative supplies a reason it was fine. Set beforehand, it is a decision you have to actually make.

The rules he will give you, and the ones he will not

Darius refuses to hand out over-specific tactical advice. He compares it to someone describing how they crossed a battlefield: I ran three steps and then I dived. It worked for them, on that day, on that field.

What he does repeat are what he calls the biblical rules, both from Paul Graham. Talk to customers. Do things that do not scale.

Two of his own sit alongside them. Test the ugliest version, because if people will not pay for it when it is ugly they will not pay for it when it is better. And ship earlier than feels comfortable.

He is clear-eyed about where this leads. The Founders it helps most, he says, are the ones whose ideas it kills.