Cüneyt Mertayak
← All articles

Building software

On Leaving the Customer’s Problem Too Soon

Water drips from a mossy ledge into a smooth hollow worn into the stone below.

If you're trying to build a software business and you're not obsessed with the problem your customers face, you're f'ed.

Say a customer tries your first version and goes back to the spreadsheet they've used for years. I'd want to sit down with them and see what happens the next time they open that sheet. What does it let them do that my product missed?

The shiny object problem used to be easier to spot. You'd leave a project half-built when a new idea caught your attention. With coding agents, you can finish a project every week and move straight on to the next one. By the time a few people try last week's release, you're busy with this week's, and you may never find out whether it helped them.

The ability to finish another project can make it easier to leave the previous question unanswered.

A new project offers a fresh start

Early on, progress is easy to see. There was no page; now there is a page. A button works. Data arrives. You have something to show.

Later, the questions get less comfortable. Who needs this? How will they find it? Why did someone stop halfway through? Is the problem important enough for them to change what they already do?

A coding agent can help build a simpler flow or add the measurements needed to investigate. Its output still needs checking, and it cannot supply the missing experience of someone actually using the product. That takes getting the thing in front of people and following up.

Meanwhile, the next idea offers another stretch of visible progress. I think that makes switching projects especially tempting now: you can leave an uncertain result behind and have a new working demo by the weekend.

Keep the problem, reconsider the solution

My personal reading pipeline started with wanting to turn videos into something I could read offline, on printed pages like a newspaper. From individual videos, it grew into Kindle reading, automatic fetching from YouTube channels, and reading on the web and my phone. I also added ways to send material from the platforms I used.

Those changes came from using it. The underlying need stayed recognizable: I wanted to read material that otherwise required more watching and listening than I wanted to do. That gave me a reason to return to the same project as the way I used it changed.

This is what I mean by staying with the problem. A particular feature, interface, or technology can turn out to be wrong. Caring about the problem gives you a reason to change those things. It doesn't require defending your first solution forever.

For a personal tool, your own repeated use gives you somewhere to start. For something intended for other people, your experience alone won't answer their questions.

Getting it used is part of the work

If you want other people to benefit, reaching them belongs in the plan. It is easy to leave that until the product feels ready, then discover that the next feature is more appealing than finding someone to try what already exists.

For a reading app, I would want to know whether someone gets from saving a video to reading its summary. If they never open it, another capture option may do very little. Perhaps they forgot it was there. Perhaps the summary wasn't useful. Those possibilities lead to different changes, and I need to investigate before choosing one.

Enough exposure means enough relevant use to learn something. A few people with the problem may give you more to work with than a large audience admiring a screenshot. Watch what they do, ask about what happened, and give them a reason to try the next version.

Staying also has limits. If repeated use and conversations show that the need is weak, changing direction can be sensible. The question is whether you learned something that changed your mind, or simply found something newer to build.

Before starting the next thing

Before opening another project, I would ask: Have I learned enough from people using this to move on, or have I just reached the part that coding can't finish for me?