8. Options before choices

§The question this chapter answers: what are all the ways this problem could be solved, and which two or three deserve to be tested?

§The problem statement is written. The pull toward the solution you had in mind on day one is now at its strongest, and this chapter exists to resist it for a short while.

§ 8.1Three rules of divergence#

§No idea is precious. Including the one you founded the company on. If it is good, it will survive the comparison. If it will not survive the comparison, you want to know now.

§Quantity first. Ideas improve when there are many of them, because the tenth idea is only reachable through the first nine. Write them alone and in silence before anyone speaks; the fastest way to get one idea from five people is to let them talk.

§Steal from adjacent industries. Every problem has been solved somewhere, for someone else. How does aviation handle a checklist? How does a hospital handle a handoff? What would a game studio do with your onboarding? Jake Knapp's sprint calls this lightning demos: bring real screenshots of real products that solved a piece of the problem, and say exactly what to take from each. A reference without an instruction is decoration.

§ 8.2Where to look#

§Adjacent industries, as above. Different markets: how is this problem solved in a country with different infrastructure, or for a customer ten times larger or smaller? Nature, occasionally, if the problem is about flow or allocation. History: how was this handled before software, and what did the manual version get right that the software lost?

§And the general-purpose tool. In the AI era the most important adjacent solution is often a chat window and a saved prompt. Your customer may already be solving a version of the problem with one. That is a solution on your list, and it is the one you have to beat.

§ 8.3Clustering back to the problem#

§When the ideas exist, group them. Not by how they look, by which part of the problem statement they attack. Some ideas fix the workaround, some remove the situation, some reduce the impact. Name each cluster, and for each ask: which line of the five-line statement does this change? An idea that changes no line is a feature looking for a problem, and there will be several. Set them aside without regret.

§Then rank the clusters on four things: how feasible with the people you have; how much of the impact it removes; how quickly it could reach a customer; and what it would need that you do not have. The ranking is not the decision. It is the shortlist.

§ 8.4The solution canvas#

§For the two or three strongest clusters, one page each.

§Core concept. One sentence. The key differences from what exists. The main assumptions it rests on, written as claims that could be false. User journey. How the person discovers it, starts using it, gets value from it, and tells someone else. Four lines. If the fourth line is blank, ask why anyone would. What it requires. The must-haves, the nice-to-haves you will not build first, the hard parts, the resources. Fit. Who exactly it is for, how it addresses the problem statement line by line, why now, and who else is close. Risks. Technical, market, business model. Marty Cagan's four risks apply here at the level of a single option.

§The canvas is where a clever idea meets its assumptions. Many clever ideas do not survive the second box.

§ 8.5What you leave with#

§Two or three solution canvases, each one page, each with its assumptions written as claims that could be false. A list of the ideas you set aside and the line of the problem statement each failed to touch. And a sentence, for each canvas, naming the single assumption that, if wrong, kills it. That sentence is what the next four chapters test.