During a training session, a square appears in different positions on the screen while letters are spoken aloud. The user needs to follow two sequences at once. If the square’s current position matches its previous position, press one button. If the letter repeats, press the other. Sometimes both signals match, and sometimes neither does.

When you spend a long time developing a product like this, its interface starts to seem obvious. You know what every button does, where to look, and when to press it. That’s why your first encounter with a real user quickly brings you back down to earth.

Version one: they’ll figure it out as they go

Initially, the tutorial was a short demonstration session. It showed the basic mechanics of N-Back but barely explained the controls.

I expected people to see the interface and figure out what to do next. In practice, some users couldn’t even work out how to start a training session. Others started their first session, pressed the wrong buttons a few times, and left.

Some people closed the page while still going through the tutorial. I decided the problem was unclear rules. That may have been the case, but I couldn’t prove it: my analytics showed users leaving but didn’t explain why.

Version two: now I’ll explain absolutely everything

My next attempt was the “Academy” — a separate section with theory, examples, and explanations of the controls.

The logic seemed reasonable: first, people take their time learning the rules, then move on to practice already prepared. But it ended up taking too long. Users skimmed through the theory and often never reached the point where they could finally try the exercise themselves.

I tried to solve the problem of an unclear tutorial with even more tutorial content. I think this was about when I started suspecting I was heading in the wrong direction.

Version three: action first, explanation second

After that, I decided to drop the long introduction and run the tutorial directly on the actual playing field.

First, the app shows only the square and teaches users to notice when its position repeats. Then the square is temporarily disabled, and the user works with sound alone. After that, both types of signals are combined.

A hint doesn’t just tell you which button to press. The app presents a predefined sequence of signals, knows the correct answer at each step, and waits for the user to take the required action. This means people aren’t reading instructions separately from the interface — they immediately try following them in practice.

The first steps use 1-Back mode (easy level): each new signal only needs to be compared with the previous one. This is the simplest version of the exercise. Once the hints end, an independent test consisting of 5 signals begins. At this point, the interface no longer highlights anything: users must determine for themselves whether the square’s position, the sound, both signals, or neither has repeated.
I like this version more than the previous ones. It gets people to their first meaningful action faster. But “I like it” still doesn’t mean “users understand it better.”

I opened Metrica to check my assumptions

To evaluate the new tutorial, I opened Yandex Metrica and looked at two goals: “Tutorial started” and “Tutorial completed.”

Over the week of September 8–14, 57 visitors started the tutorial, and completion was recorded for 26. But these numbers still don’t answer my main question: did it become clearer to people?

The goals themselves are configured correctly. The first triggers when the tutorial starts. The second triggers when someone has gone through the hints and finished the independent five-signal test.

But there’s a problem: finishing the test doesn’t mean completing it correctly. As the goals are currently configured, even if a user gets almost everything wrong, Metrica will still record the tutorial as completed. It turns out I measured tutorial completion, not an understanding of the rules.

My takeaways and what I’ll do next

I won’t delete the existing goals. They show how many people started the tutorial and how many reached the end.

I’ll also create a multi-step goal that tracks the user’s journey in order:

• Started the tutorial;

• Reached the end of the independent test;

• Started their first full training session;

• Finished it.

This goal will only trigger if someone completes all the stages in the correct order within a single visit.

I’ll add a separate event for those who actually pass the test without hints. Right now, Metrica doesn’t distinguish between completing it correctly and simply reaching the last screen.

My main takeaway is a little obvious: first, you need to define exactly what counts as success, and only then configure your analytics. Regular goals show individual actions; a multi-step goal shows the entire user journey. But even a multi-step goal will only tell you that someone went through all the stages. To understand whether they grasped the rules, you need to define and track a separate metric.

Here is the current version of the tutorial. If you give it a try, please let me know: was there a moment when you didn't understand what to do next? And what exactly was confusing at that point?

If you want to see what I ended up with first, try my trainer.

How Multitasking Became a Problem

In my work, I constantly switch between different tasks and often work in multitasking mode. Strangely enough, AI tools have only made this problem worse for me: they allow me to run more processes in parallel, making it even harder to keep my attention on a single task.

Why Dual N-Back?

I began looking for exercises that could help me train my working memory and concentration. That is how I discovered Dual N-Back and started reading the research around it.

The results of these studies are mixed, so I did not treat the method as a magical way to increase intelligence. I simply wanted to find a regular and measurable exercise and test it through my own experience.

Why Existing Trainers Did Not Work for Me

I then explored the existing trainers. Each of them had interesting ideas, but none completely satisfied me. Some had interfaces I did not like, some used scoring systems that were difficult to understand, and others lacked adaptive difficulty, an offline mode, or the ability to train without creating an account.

In the end, I decided to build my own trainer:

Try the current version of Dual N-Back for free and without signing up.

What I Built and What Is Still Ahead

I combined the things that were personally important to me: a calm, modern interface, clear performance metrics, adaptive difficulty, a local training history, and the ability to use the trainer without mandatory registration.

The trainer is currently completely free, supports multiple languages, and its core features are available without an internet connection. Training results are stored locally on the device. I am also working separately on optional synchronization between devices.

I originally built this project for myself and expected almost nothing from it. That is why I was surprised when other people started using the trainer, and some began returning to their training sessions regularly. It is especially rewarding to see people continue practicing and gradually improve their results.

If you decide to try the trainer, I would be glad to receive your feedback. I am especially interested in whether the first-run experience is clear, whether the scoring system is easy to understand, whether the difficulty feels right, and what you think is missing.

I continue to actively develop the project and pay close attention to thoughtful suggestions.

Try the trainer for free without signing up and share your impressions →