How to Choose a Startup Problem Worth Ten Years

How to Choose a Problem Worth Ten Years

A startup problem is worth ten years when consequence, real demand, founder advantage, structural timing, plausible economics, ethical value, and personal staying power survive the same test. Commitment alone is not evidence. Market evidence alone is not a reason to spend a decade.

The decision should end in a written memo that states why this problem matters, what has changed, what the founder uniquely knows, what must be true, how those claims will be tested, and which results should cause the team to stop.

Ten years is a filter, not a forecast

In episode 99, Andy Ellwood says he developed four non-negotiables before becoming a founder again. The criterion he explains most fully is that the next company had to address a problem he wanted to think about, discuss, and pursue for ten years.

Grocery transparency passed the test because Ellwood had already built Basket and continued returning to the problem after that company ended. He says changes in grocery data made a second attempt feel different enough to justify Stretch.

The ten-year horizon filters out ideas that are attractive mainly because they are easy to launch. It does not predict that the company will last ten years, raise capital, find demand, or produce a good outcome. The point is to test the founder's willingness before the work becomes expensive and social pressure makes stopping harder.

flowchart TD
    A["Problem attracts attention"] --> B["Consequence and people affected"]
    B --> C["Observed behavior and demand"]
    C --> D["Founder knowledge and access"]
    D --> E["Structural change and timing"]
    E --> F["Economics and distribution"]
    F --> G["Ethics and second-order effects"]
    G --> H["Life fit and ten-year willingness"]
    H --> I["Commitment memo and kill criteria"]
    I --> J["Run the next falsifiable test"]

The process ends with a test, not a declaration.

Start with consequence, not market size

The Founders Fund essay “Choose Good Quests” argues that people with unusual skills and resources should work on consequential problems rather than defaulting to easy, status-preserving projects. It also introduces a “player card” idea: a person's resources, skills, network, and constraints affect which hard problem is a credible fit.

The essay is an argument about responsibility and ambition, not an empirical startup-selection model. Its useful question is whether success would make a meaningful part of the future better. Its weakness is that consequence can become theater when the beneficiary, mechanism, and tradeoffs remain vague.

Write the people affected, the current harm or burden, the behavior that demonstrates it, and the result that would count as improvement. “Fix food affordability” is too broad. “Help a defined household compare its recurring basket across feasible stores with current prices and explicit tradeoffs” can be observed and tested.

Replace the painkiller metaphor with demand evidence

Ellwood uses the familiar distinction between a painkiller and a multivitamin. A person may forget a vitamin and continue the day, but a severe headache can interrupt work and create immediate willingness to act.

The analogy is useful when it forces specificity. Who is in pain? What do they do now? How much time, money, risk, or frustration do they accept to solve it? What event makes the problem urgent?

It becomes misleading when treated as a universal law. Preventive medicine, infrastructure, security, climate adaptation, education, and compliance can matter before a user feels immediate pain. Some urgent problems still lack viable economics. Some delightful products create durable value without relieving distress.

Use the metaphor as a prompt, then collect behavioral evidence. Interview people about recent actions rather than hypothetical enthusiasm. Observe workarounds, abandoned attempts, repeated purchases, budget authority, switching costs, and the point where the current solution fails.

Steve Blank's customer-development guidance treats a startup's early beliefs as hypotheses to test. It warns against turning customer conversations into requests for a feature list. The stronger question is what a person is trying to accomplish, how the person behaves now, and whether a minimum solution changes that behavior.

Founder advantage is earned access to the problem

Founder fit is often described as passion or identity. A more useful definition is access to evidence and action. Does the founder understand the workflow, language, incentives, regulation, data, buyers, and failure modes? Can the team reach users and partners who will test the claim?

Ellwood's Basket experience did not guarantee that Stretch would work. It gave him historical knowledge, relationships, and a record of the unresolved problem. It also created risks. A repeat founder can overfit to an earlier market, defend a familiar solution, or interpret persistence as proof.

The NBER working paper “Age and High-Growth Entrepreneurship” studied employer startups using US administrative data. Among its findings, prior experience in the specific industry predicted higher rates of upper-tail growth. That is population-level evidence about association, not a guarantee for one founder or a method for choosing one idea.

Treat domain experience as an advantage to test. Write what the team can see or do that a capable outsider cannot yet see or do. Then ask what prior belief the new evidence might overturn.

Timing needs a structural change

“The market is ready now” should resolve into a mechanism. A new data source, regulation, cost curve, interface, distribution channel, behavior, or technical capability may change what is possible. Fashion and investor attention can create activity without durable demand.

For a second attempt, name what was impossible, uneconomic, inaccessible, or unacceptable the first time. Then show the current evidence. Ellwood attributes Stretch's timing partly to pandemic-era changes in grocery data. A diligence process would need to identify the actual data rights, coverage, latency, economics, and dependencies behind that statement.

A structural change can also make a problem worse. Personalization may improve relevance while expanding surveillance. Automation may lower cost while reducing accountability. The timing section should include both the new capability and the new risk.

Economics and distribution belong in the first memo

A consequential problem does not automatically produce a sustainable company. The memo should identify who uses the product, who pays, why the payer benefits, how the product reaches users, what it costs to deliver, and which dependency can change the margin.

For consumer comparison, the business model may include referral commissions, retailer partnerships, advertising, subscriptions, or data services. Each one can create different incentives around rankings, coverage, and privacy. The choice belongs in the problem decision because it can change what the product optimizes.

Do not force a detailed ten-year financial model onto uncertain evidence. Do require a coherent path from user value to payer value and a test that can invalidate it.

Ethics is part of product feasibility

Ask who could be harmed if the product succeeds exactly as designed. A price-personalization tool may increase retailer revenue while making essential goods less transparent. A household assistant may save time while creating an intimate purchase and location record. A marketplace may expand choice while steering people toward commercial partners.

Ethical analysis is not a final review added after product-market fit. Trust, consent, legal exposure, and partner behavior can determine whether adoption is possible. Name data that should not be used, outcomes that should not be optimized, people who need protection, and governance that must exist before scale.

The founder should also ask whether the claimed good survives the business model. A company cannot rely on a mission statement to resolve incentives built into ranking, pricing, or advertising.

Personal staying power includes the rest of life

Ten-year willingness is not the same as endless sacrifice. Ellwood says the founder decision was made with his wife and family. That detail belongs in the framework because a long commitment consumes time, attention, financial risk, location flexibility, and emotional capacity.

Write the conditions under which the work is sustainable. Include compensation, health, caregiving, geography, travel, co-founder expectations, and the possibility that the right decision is to work on the problem without founding a venture-backed company.

A founder who ignores those constraints does not become more committed. The plan becomes less truthful.

Write one commitment memo

Schedule a review before enthusiasm or sunk cost can rewrite the criteria. The kill conditions should include both company failure and mission failure. A growing product that creates the wrong outcome may be evidence to stop or redesign.

Choose the next test, not the final identity

The goal of the framework is not to prove that a founder has found a calling. It is to select the next expensive assumption and test it with the least irreversible commitment.

Ellwood's return to grocery technology is useful because it contains both continuity and changed conditions. The problem held his attention, but the second attempt still had to confront new data, business-model, product, and trust questions.

Read [[Andy Ellwood on Building Stretch After Basket]] for that story. [[E056 Content Plan|Episode 56]] develops founder motivation and idea evaluation as a related cluster. [[From Selling Hours to Selling Outcomes]] offers another version of the same discipline: define the result before designing the operating model around it.

A ten-year problem is not one you promise never to leave. It is one important enough to examine seriously, familiar enough to approach with earned insight, and testable enough to abandon when the evidence no longer supports the quest.

Sources and editorial notes

This framework uses the episode 99 transcript, the Founders Fund “Choose Good Quests” essay, Steve Blank's customer-development guidance, and the NBER working paper “Age and High-Growth Entrepreneurship.”

The frameworks are decision prompts, not validated predictors of startup success. The NBER result is a population-level association and does not guarantee an individual outcome. AI assisted with research organization and drafting under editorial review. Publication remains unauthorized.