You’re Building for Other People, Not Yourself. Right?
By Leo
I have vibe coded quite a few projects now, and honestly, it is awesome. Things happen so fast. You describe something, see it working, and immediately want to build the next thing. It feels like you are getting a lot done.
And maybe you are. But that does not necessarily mean you are building anything your customers want or need. That is really my whole point.
When implementing an idea becomes easy, it is tempting to stop asking whether the idea belongs in the product. Feedback from people who might actually use it is what brings that question back.
Just because you can build it does not mean you should
You come up with a new feature every day and start implementing it straight away because you can. A settings page becomes a dashboard. The dashboard needs an assistant. The assistant needs another way to customize everything.
None of those ideas has to be bad. The problem is that you have no idea whether anyone needed them in the first place.
If you are creating for fun, go ahead. Experimenting is part of what makes this exciting. But if your goal is to sell a useful product, your own enthusiasm cannot be the only thing deciding what gets built.
Before you start the next feature, ask which user problem it solves and what makes you think that problem matters. Sometimes you will have a good answer. Sometimes you will realize you just wanted to build something new.
Showing friends can teach you a lot
For me, showing friends has been one of the most useful ways to get feedback, especially when they are a reasonable match for the people I want to reach.
You get their thoughts while they are using it. You see where navigation feels unclear, which wording makes them hesitate and whether they understand what a feature is supposed to do. You can ask a follow-up question immediately instead of guessing what a short comment meant.
If you are not doing this, try it. Give them a task and let them attempt it before explaining the interface. It is very easy to accidentally guide somebody through the exact problem you needed to notice.
And write down what they say. You do not have to act on every point, but you should at least take it into consideration.
The comments you dismiss are worth keeping
Somebody will point out a problem that does not feel like a problem to you. You know where the button is. You understand the phrase. You would never use the feature that way.
Keep the observation anyway. When the next relevant person tries the same task, see what happens before asking about the earlier comment. If they independently run into the same difficulty, that is much more useful than persuading them to agree with your friend.
Repeated confusion can show you something you did not notice because you already know the product too well. That does not automatically tell you the right fix, but it gives you a reason to investigate.
You are building for other people, not yourself. Right?
Your friends did not sign up to be your research team
There is a catch. Asking friends for feedback can get annoying for them. I have shown projects to friends who clearly were not interested. That is uncomfortable for both sides, and the feedback usually is not very useful either.
Someone can care about you without caring about an invoicing flow or the third version of your onboarding. A polite “looks good” may mean they want to be supportive, not that you have solved the problem.
Ask whether they have time. Be clear about what you need. And if they are not interested or are not your audience, find someone else.
Finding feedback online has its own cost
That is why I became more interested in getting feedback online. I tried posting on X and Reddit. Those posts could bring people in, but getting visitors to come back with detailed feedback was another matter.
I also tried messaging people and offering to test their product if they tested mine. That worked better for me, but it took a lot of time. You still have to find the person, explain the exchange and do your half of it.
I kept coming back to the same conclusion: nothing is really free. Friends may help because that is what friends do. A stranger usually needs another reason to spend time carefully trying your product and writing useful observations.
Give people something worthwhile in return
These days, I am much more willing to compensate people for that work. Cash is one option. Access to the product can also make sense if it is something they want.
That second option is particularly interesting to me. If a person chooses your subscription as their reward, they have shown an interest in the product itself. You still need to check whether they match your audience, but you have a much better question to start with: what do they hope your product will help them do?
This is part of what I am building Testbay around. Developers can offer a defined reward, set a focused task and collect feedback in one place. Through the MCP integration, a coding assistant can help prepare a draft and read the feedback afterward. You still review, fund and publish the test yourself.
The aim is to make human feedback a repeatable part of development, rather than something you only remember to ask for when a launch disappoints you.
Use the speed to fix the right things
AI makes it easier to act on what you learn. That is a good thing. A confusing phrase or unnecessary step does not have to stay there for weeks while you wait for development time.
But the speed only helps if you know what needs changing. A request for a new feature is worth understanding before you implement it. Sometimes the person needs a different outcome. Sometimes the existing feature is simply too hard to find. Sometimes removing a step is the best improvement you can make.
Before building something else, give a relevant person a chance to use what is already there. Listen, write it down and decide what deserves your attention.
Then use all that development speed on something that matters to them.
Put the feedback to work
Choose one product question, write a focused brief and offer people a reason to help you answer it.
Create a brief