How Feedback From Users Shaped the Onboarding of Detox Mental

I learned an idea from professional writers that applies extremely well to business and product building: as a creator, you must learn to “kill your darlings.”

“Darlings” in this context are those ideas that the creator is emotionally invested in before testing them. The ones that he or she thinks will work even before real users, customers, readers, or any final consumer has interacted with them.

Many ideas feel great before they are tested against reality and end up being a failure when shipped, but since we are emotionally attached, learning to let them go is not always as easy as it sounds.

Here’s my own story with such an idea in recent times:

Detox Mental was born from a particularly successful article wrote in 2021. Its KPIs—reads, comments and shares—demonstrated that many people read it until the end, even when the approximate reading time was 45 minutes (~8.000 words).

And not only that. People finished it and wanted more.

The comments were full of readers asking for more information and further advice on their particular situation, so based on this, I built the Detox Mental course, consisting of 15 audio sessions expanding the ideas of the article.

At this point, the next step was to redirect users from the article to the course.

Since I “knew” that many people got to the end, I thought the article itself was the perfect funnel: although many would exit the app before finishing it, the genuinely interested users will stay and be hungry for even more content.

And so I decided to put a single entry point to the course at the very end of a 8.000-word article, against the advise from a few users that reviewed the app before it launched.

Of course, the result was what everyone—but me—expected: the number of people reaching the course was low.

Based on this, it was time to kill my own “darling.” I was disappointed that the article alone did not work as expected, and it was hard to let go of my dream that it would bring users to the course without any further effort, but numbers showed an undeniable reality.

I had to let go of my wishful thinking and accept that a shorter introduction to the product was needed. Something that demonstrated the value of the app in only a few minutes.

And so the first iteration of the Detox Mental onboarding was born: an AI-assisted chat where new users could learn the basic ideas of the app, have a first writing experience, and get curious about the article and the course.

This worked great—as long as I was physically there to explain it myself.

When doing in-person demos of the app, users were intrigued by the onboarding experience.

“This is really cool!” They’d say.

But again, numbers showed that “really cool” wasn’t enough.

The full onboarding took ~5 minutes to be completed, and who has five minutes for a new app in 2026?

Further feedback was received from new users. They said “I liked the idea, but when I got to the app, I faced a long onboarding and unfortunately decided to get out.”

And so the second iteration was born: a much shorter onboarding that is completed in less than 30 seconds.

Official numbers are still to be confirmed, since Detox is currently being tested only with a small subset of trusted users.

However, the trend is clear: from “I couldn’t get to the end” on the first iterations, users are now reporting that they know what the app is about in less than a minute.

To do this, the idea that the article alone would make the app successful had to die. It was tested and it didn’t work, so the product needed to move forward. 

Hunches and intuition are very valuable, but if they do not hold against reality, creators must be ruthless in moving on and listening to the real signal: the market.

One must be willing to listen to reality in order to build successful projects. If numbers and conversations with users confirm that an idea does not work, the skill to let it go and iterate fast is not only a nice-to-have—it’s absolutely necessary.