How to build a product that achieves product-market fit from day one

Executive overview

Most founders treat product-market fit as a milestone to reach. It can arrive without you noticing — Granola had it on day one and took six months to realise it.

The solution is not better intuition upfront. It is a tight feedback loop: probe the market, observe what happens, adjust fast. The right product shape is unknowable until it makes contact with the world.

Talk to users constantly — but don't just do what they say

• Talking to users is the lifeblood of building product; without it you will not build something really good.
• Lower the activation energy to get feedback — if it's hard, it won't happen consistently.
• Systematise access: Granola runs standing user interviews four days a week that anyone at the company can join.
• One clear signal from one user is enough to act — you don't need 10 people to confirm an obvious problem.
• Never ask users if they would use something; ask what they would do next, then probe negatively.
• Ignore all positive feedback in interviews; discount praise by default.
• Don't build a prioritised list of user requests — users who asked for features often don't use them.

Explore before you exploit

• Spend the pre-launch period in explore mode: try many shapes, change the product radically, cut freely.
• Granola cut ~50% of what it had built before launch — impossible to do post-launch without backlash.
• Once you know the right shape, switch to exploit mode: polish relentlessly, but only the right shape.
• Polishing the wrong shape wastes everything.
• The hardest question is knowing when to stop exploring — there is no reliable signal; launch when you'll learn faster from real users than from controlled tests.

Use conservative metrics to stay honest

• It's easy to delude yourself with vanity metrics.
• Granola's definition of an active user: completed at least one new meeting that day with over five minutes of transcription.
• Opening the app or reviewing old meetings does not count.
• Conservative metrics prevent rationalising away bad results.

Build products with soul

• A product with soul feels coherent and consistent — every screen comes from the same values.
• The opposite: a product where you can infer the company's org chart from its design.
• As teams grow, alignment on values must deepen; misalignment shows up in inconsistent product feel.
• Think of your product as a person your users build a mental model of — unexpected behaviour breaks trust.

Augment humans, don't replace them

• Granola's direction: augment what the user does, not replace them.
• Starting with meeting notes, expanding to follow-up emails, memos, meeting prep, and cross-meeting analysis.
• The goal is to make the user more capable, not redundant.