Please Stop Building Features for Five Minutes...
Summary: Alexandra Renée Poelstra discusses the tendency in startups to prioritize building new features over understanding real customer needs. They highlight the importance of listening to customers rather than relying solely on a product roadmap, emphasizing that survival in startups often depends more on meeting true customer demands than merely adding features. They caution against the danger of moving too quickly in the wrong direction and suggest practical ways, such as effective questioning and understanding customer behavior, to achieve product-market fit. The post encourages startups to focus on understanding problems and assumptions rather than hastily building features.
There is a very specific type of dopamine that comes from building a new feature.
You have an idea. You can picture it. You start sketching it out. Someone says, “Ohhh, that would be cool.” Suddenly it’s on the roadmap, your developer is questioning their life choices, and three weeks later you have added a button that approximately seven people will ever click.
Welcome to startups.
I say this with love because I am extremely susceptible to the shiny-new-feature problem. Building feels productive. It’s tangible. You can point at something and say, “Look! Progress!”
Talking to customers is messier. They say confusing things. They contradict each other. Sometimes they tell you the feature you are ridiculously excited about is…fine.
Rude.
But here’s the thing: your customers are probably more valuable to your roadmap than your roadmap is.
Startups have a survival problem
According to the U.S. Bureau of Labor Statistics, 770,523 private-sector establishments were part of the cohort that opened in the year ending March 2019. Five years later, in March 2024, 51.5% were still operating. By March 2025, that number had fallen to 46.8%. (bls.gov)
That doesn’t mean half of startups disappear because somebody built the wrong dashboard. Business failure is complicated. Cash, competition, timing, team dynamics, regulation, pricing and plenty of other things can take a company down. CB Insights’ 2026 analysis of more than 400 startup post-mortems found recurring patterns across failed companies rather than one magical, singular cause. (cbinsights.com)
But there is one thing founders can actually do something about surprisingly early: figure out whether people care.
Not whether they think your idea sounds cool. Not whether your friends tell you they would “totally use that.” Not whether 143 people liked your launch post. Care as in: I have this problem. It bothers me. I am already trying to solve it. Your thing makes that easier. Where do I sign up?
Very different energy.
We have made “building” synonymous with progress
This is one of the stranger parts of startup culture. We are constantly told to build faster, ship faster, launch faster and iterate faster. There is absolutely value in speed, but speed in the wrong direction is just a very efficient way to get lost.
And now AI has made this even more interesting because we can build things at a pace that would have seemed mildly unhinged a few years ago. That is amazing. It is also dangerous.
Because the cost of building the wrong thing has dropped dramatically.
You can now take a mediocre assumption and turn it into a beautifully functioning mediocre assumption by Tuesday.
Congratulations?
The scarce resource isn’t always development anymore. Sometimes it’s understanding.
Product-market fit is not a vibe
Founders talk about product-market fit constantly, but it can become one of those startup phrases we say so often that it starts losing meaning. There are actual ways to look for it.
One well-known framework comes from growth consultant Sean Ellis. After benchmarking roughly 100 startups, Ellis found that companies showing strong traction tended to have at least 40% of surveyed users say they would be “very disappointed” if they could no longer use the product. Companies struggling to gain traction generally fell below that mark. (review.firstround.com)
Think about that question for a second. It’s not, “Do you like our product?” or “Would you recommend us?” or even, “Is this useful?”
It’s: How disappointed would you be if this disappeared tomorrow?
That’s a much tougher question. And frankly, we need more tough questions.
Talk to humans before you build for humans
One of my favorite examples comes from Productboard.
Before the company became what it is today, its founders spent about a year doing customer discovery and prototyping. According to First Round Review, they conducted more than 1,000 customer interviews and tested 13 different prototypes while figuring out what the product needed to become. (review.firstround.com)
One thousand.
I am not suggesting you cancel the next six months of your life and begin wandering the earth interviewing people like some sort of startup anthropologist. But maybe talk to five. Then five more. Especially before committing meaningful time or money to something.
And when you do, stop pitching.
This is the hard part. Founders are salespeople by necessity. We explain. We contextualize. We persuade. Someone gives us a tiny objection and our brain immediately goes, Oh! That’s only because you don’t understand. Let me explain this differently.
No.
Zip it.
You are not trying to win the interview. You are trying to learn something.
Ask better questions
The quality of the answers you get is usually connected to the quality of the questions you ask. Instead of asking, “Would you use a platform that helps you do X?” ask, “How do you handle X today?” Instead of, “Would this feature be helpful?” try, “What is the most frustrating part of doing this now?”
And instead of asking someone whether they would pay $29 a month for something, ask whether they are paying for anything to solve that problem today.
One of my favorite questions is simply: “Tell me about the last time this happened.”
That question is sneaky-good because it takes people out of hypothetical land. Humans are wonderfully optimistic in hypothetical land.
Yes, I would absolutely exercise five mornings a week. Yes, I would meal prep every Sunday. Yes, I would use your financial planning app daily.
Hypothetical Me has her life together.
Actual Me has eaten cheese over the sink.
Your startup needs to build for Actual Me.
Pay attention to behavior more than compliments
This may be the hardest lesson because people are nice. Especially when they know you’re building something.
They will tell you, “This is awesome.” They will say, “I love this idea,” and “You guys are going to crush it.”
And then they will never use it again.
This doesn’t make them dishonest. It makes them human.
So watch behavior. Did they come back? Did they invite someone? Did they finish onboarding? Did they actually use the feature? Did they pay? Did they complain when something stopped working?
That last one sounds negative, but sometimes a complaint is weirdly encouraging. Nobody gets angry when a product they don’t care about goes offline.
They just leave.
Don’t build everything your customers request either
And here comes the plot twist: listening to customers does not mean turning your roadmap into a customer suggestion box.
If ten customers request ten different features and you build all ten, congratulations—you have invented enterprise software.
Your job is to listen for the problem underneath the request.
A customer might say, “I need an export button.” Maybe they do. Or maybe what they’re really saying is, “I need to show this information to my boss.”
Those are different problems.
One leads to an export button. The other could lead to reports, dashboards, sharing, permissions, automated emails or something you haven’t thought of yet.
Customers are remarkably good at explaining their problems. They are not always responsible for designing your product.
That’s still your job.
Give yourself a “prove it” rule
Before something substantial goes onto your roadmap, make the idea earn its way there. You don’t need a 47-page product requirements document. You just need to understand a few basic things.
Who actually wants this? What problem does it solve? How are people solving that problem now? How often have you heard about it? What happens if you don’t build it? And is there a smaller, cheaper way to test the idea first?
If the answer is basically, “Well, nobody specifically asked for it, but I think it would be really cool…”
Fantastic.
Put it in the idea pile.
The idea pile is lovely. Things can live there. They have snacks.
Build less. Learn more.
Startups are fundamentally giant collections of assumptions. We assume we understand the customer. We assume the problem matters. We assume our solution is better. We assume they’ll pay. We assume we know where to find them.
Your job isn’t to magically be right about all of those things on Day One. Your job is to figure out which assumptions are wrong before they become expensive.
And there is something wonderfully freeing about that.
You don’t need to have every answer. You need to stay curious long enough to find them.
So yes, ship things. Experiment. Build the weird prototype. Try the idea everyone thinks is slightly nuts. That is part of the fun.
But every once in a while, close Figma. Close your project management software. Step away from the roadmap.
And go talk to a human.
They have a funny habit of telling you what to build next.
.png?_a=BAVMn6DY0){kind=small}