2 truths and a UX startup lie

We're playing a game this month.

Three statements about UX in early-stage startups. Two are true. One is something almost every founder believes, and it's costing them.

Play with us, see if you can spot the lie before the reveal.

Statement 1: Talking to five users is enough to uncover most of your UX problems.

Statement 2: Your users can't tell you what to build.

Statement 3: Good UX is a post-PMF investment – you sort the experience once you know the product works.

Got your answer? Here's the reveal.

The lie: Good UX is a post-PMF investment

"We'll sort the design once we've validated the idea" is probably the most expensive sentence in early-stage startups.

You're pre-PMF, runway is finite, and UX feels like polish – something you layer on top once the core product is working. Ship first, design later.

Sadly, the economics run the other way.

The cost of fixing a UX problem post-launch is up to 100 times more than fixing it during the design phase. 

Here's how that compounding works. 

1. An onboarding flow that loses 40% of users doesn't just cost you those users. It costs engineering time to rebuild it on top of an architecture now optimised for the wrong experience. 
2. A component pattern that's slightly wrong doesn't stay contained – it replicates through your codebase until a three-hour fix becomes a three-week refactor. 

An even deeper problem is that "design it later" misunderstands what UX truly is. 

UX isn't the visual layer. It's the decisions about how your product works: what the flows are, where the friction lives, what happens when something goes wrong. 

Make those decisions badly under pressure and you haven't saved time. You've borrowed it at a very high interest rate.

Truth 1: Five users is enough, for the right kind of question

Five users in a qualitative usability test uncover approximately 85% of your core problems. After five, you're mostly seeing the same things again.

Five 30-minute sessions is two and a half hours of your time. The research that gets skipped is the research founders think is too small to matter.

Truth 2: Users can't tell you what to build, but they're exceptional witnesses to their own friction

Steve Jobs said "people don't know what they want until you show it to them." Founders use it as a licence to build without looking up. They're misreading him.

Jobs was arguing against asking users to imagine products that don't exist – focus groups, feature wish lists. He was right about that. But he was an obsessive observer. He watched what people struggled with, that's the research he was doing.

Users are unreliable narrators of solutions. They'll describe a slightly better version of what they already have. 

But they're exceptional witnesses to their own friction. The question that changes a session isn't "what would you like this to do?" It's "show me what you'd do next." Then stop talking and watch.