CTO in a startup. Problems and conclusions
Important disclaimer – this is not a successful story. We did not reach the payback level, much less profit. Below I will tell in detail why and how this happened, what and at what point I did not pay attention to and what it cost in the end.
I think this article will be most useful for early stage startups, founders, and CTOs with little experience.
Also, this is my first post here, so I'd appreciate any feedback!
About the author
My name is Zhenya and I have a large technical background in large companies in the field of web development, and in startups in different roles I have been for more than 5 years. I incredibly enjoy using technology and people to effectively implement cool projects, help businesses achieve their goals and benefit people.
What kind of startup is this?
The startup we'll be talking about is ZNZN.RU – an educational platform for students with a focus on exam preparation.
The idea of the project came from my partner and we developed it together. I think that this idea appeared because when we were students, existing online resources had the following disadvantages:
It was very difficult to find a specific answer to a specific question. In the process of preparation, I had to shovel through dozens of different sites and try to collect information piece by piece.
The platform's content was of poor quality and could not be trusted.
Interfaces from the 2000s that were very difficult to use. Many resources were not optimized for viewing on mobile devices at all.
In addition, we saw the potential for such a platform to be widely adopted in higher education as a tool for teachers and students.
Overall, we wanted to make a convenient tool for exam preparation, which would have quality material, useful functionality and a well-thought-out interface. And we also wanted to somehow make money on it.
My connection stage
The idea for the project came to my partner in mid-2020. That was the first time we discussed it. I even have some notes like these.
We discussed key things and recorded them. We also thought through several possible options for the project's trajectory after the first version. We also came up with a name there. There were many options, but in the end we chose this one – ZNZN. It sounds like “knowledge”, it's unusual in appearance and sound, plus the domain was available and didn't cost much. We decided that it suited us. After all, Yandex, Google, and Dodo Pizza exist.
Since the end of winter 2021, we started preparing for the launch. We had a starting team of four people who were very interested in the project – my partner, the design lead, the creative person and me. In late spring 2021, we launched the design, a little later the development and work with content. Then I started looking for the first people for the development team.
I came in as a co-founder/CTO, fully responsible for the technical side of the project and the development team that was to be built.
The partner was responsible for organizing other teams – design, content, marketing, work with universities. And also for money, external communications and marketing.
Project life cycle
Despite the fact that we saw many potential vectors for the project's development, we understood that at the beginning we needed to focus on the core functionality and lay the foundation for future expansion. I think it was the right decision. It allowed us to test the main hypotheses without spending a lot of resources, and then move the project in a certain direction.
Also, throughout this project, the development team was formed, processes and approaches evolved. I will not go into details of working with people in this article, so as not to stretch it out, but to focus on the project. If you are interested in learning about building a development team from scratch and what mistakes I made, then please let me know in the comments and I will be happy to tell you about it in a separate article.
Development of the main functionality
This is exactly where we started.
We considered two things to be the main functionality:
the ability to upload and format questions and answers to them from the admin panel
the ability for the user to interact with them conveniently
First we made the admin panel and cool WYSIWYG editor. This was necessary so that it could be given to the content team so that they could start uploading content to the platform.
Afterwards, we implemented a user interface with basic capabilities for searching for answers on the platform and viewing them.
We made the basic functionality in 3 months, and the project was released. At that time, we had 4 subjects and 694 questions with answers on the platform.
Content is our everything
After the release of the first version, it was actively uploaded to the platform by our content team for several months. Looking ahead, I will say that the content was uploaded in the future after the release of the second version.
To work with content, they developed a separate methodology to ensure its quality.
While working with the admin panel, they gradually formulated the set of features they needed there. These requests largely formed the basis for further improvements to the admin panel in subsequent versions.
Second development cycle
The platform worked for some time, we had users, and we closely monitored them and analyzed feedback.
In particular, we realized that there is a demand for formulas, pictures and tables in questions. We realized that dividing questions by subjects is not enough. And also in the second version we wanted to implement a personal account, the functionality of notes and flash cards, so that users could not only consume content on our platform, but also work with it in convenient tools: create notes, flash cards, share them with others.
This formed the basis of the second version.
Here I would like to focus on the functionality of notes and the functionality of flash cards.
Notes
The basic hypothesis is that students need a tool to take notes online during classes. It could be more specific than Google Docs and be aimed at higher education with a thoughtful element of social interaction.
Spoiler
It turned out that it was not needed. Now I even understand why, but back then there was no particular skepticism and we made this functionality with great enthusiasm.
In that thirty second video you can understand the essence of the notes on our platform.
Implementing the notes took up quite a lot of our resources, and it turned out that users didn't need this functionality. Classic.
If I were in that situation again, when the decision was made whether to do it or not, I would make a decision only after analyzing the data that we have. Data can be obtained in different ways: custdevmarket trend analysis, competitive analysis or user feedback. But startup decisions should almost always be data driven. That's what I learned from this situation.
Flash cards
It's simpler here – these are just double-sided memory cards. They help build associations.
Unlike notes, the flash card functionality was developed as an afterthought, as an additional one. We didn't really count on it, although we understood that it could solve some of the users' problems.
Its implementation took up several times fewer resources than the notes functionality.
Here's an even shorter video about flash cards:
And what did we see after some time?
That they are used approximately 15-20 times more and more often.
It turned out that the simple ability to make flash cards and create associations for memorization is much more necessary for our users than complex and deep notes with a lot of functionality.
The second version was released at the end of 2022 with approximately this functionality.
Version 3 and OpenAI
Here we added division of content not only by subject, but also by directions and sections – this significantly improved navigation for users within the site, we noticed this in the analytics.
We also added cool integrations with OpenAI in the personal account and on the answer to the question page. Here is an example of how this integration is implemented on the content page:
I think these integrations turned out to be quite useful and appropriate. But, admittedly, we did not see much popularity of these tools among users.
I will add that in this article I have mentioned only the most key and interesting features of the project. In addition to them, a lot of other things, improvements and technical optimizations were implemented.
Also, throughout the development of the project, we paid great attention to SEO and from time to time returned to SEO optimizations – this was necessary because the content pages gradually changed, some new elements and logic were added to them, and it was necessary to ensure that new things did not have a negative impact on the platform's performance. We did it well.
What's up with the project now?
Now it is an educational platform for students, which contains almost 20,000 questions on 140 academic disciplines. We also have about 10,000 registered users on the platform and several thousand users per day.
Active development of the project is complete, we do not plan to develop anything new there in the near future. Let's see how things will be with traffic in the future.
Well, for now the project benefits people and does not bring us money 🙂 More about this in the next section.
Monetization
The most important part of any startup. Now I understand it. Unfortunately, when we started this project, our idea was very naive. We thought through several monetization vectors, but we didn’t even plan to implement them in the first versions. We thought that the ideal moment would come and then we would do them, add some killer features and get a lot of money. But at first, we were just thinking about developing functionality, about the convenience of the interface, about some cool features.
I think it was a mistake. If we had focused on business results from the very beginning, the platform positioning, target audience, and functionality could have been different.
We managed to attract funds for the implementation of this project on a competitive basis twice – however, even then we still did not direct them towards commercialization, but continued to develop the functionality that we considered necessary.
As a result, it turned out that the project is more social and charitable than commercial. Actually, what was sown is what grew. Everything is logical.
As for the monetization ideas we had, these two were the main ones:
A targeted advertising system with high conversion depending on the user's interest.
The basic hypothesis was this: due to the fact that we can understand the user's interests very well, we will be able to show him very relevant advertising offers. For example, if we see a person who studies programming issues, we can show him advertising material from a partner, for example, Yandex.Practicum or other companies potentially interested in such an audience. There were preliminary agreements with several organizations in Belarus.
According to our estimates, developing such a system with interaction with interested organizations made sense with traffic of approximately 10,000 people per day. We did not reach this figure because we relied too much on organic traffic that we would receive from search. In reality, our best figures were 4,000 people per day at the end of 2023. Then the Google search engine update came out, and the figures dropped by 2-3 times. At the moment, we receive more than 60% of traffic from Yandex.
An opportunity for teachers to teach students on our platform and earn money from it.
Something like italki, but with a focus on higher education. We also didn’t get to the implementation stage, although at some point the project received features that moved it in this direction. For example, we made some functionality for teachers that allows them to prepare materials and then conveniently share them with students. We even managed to ensure that this functionality was actually used by teachers in classes at some universities in Belarus.
I think these ideas were pretty good and had potential, but we didn't implement them because, due to inexperience, we didn't pay enough attention to the commercialization of the project.
Three main problems and conclusions
Our expectations regarding traffic volume were not realized.
And this is really a huge problem for this project, because all the monetization plans were based on the fact that there would be a lot of traffic, hundreds of thousands of users per month. We expected that we would do good SEO optimization, get a lot of traffic, and then figure out how to make money on it. We looked at competitors, saw their traffic figures and thought that if we make a user-friendly and technically high-quality site, then the traffic will come by itself.
No way, Google and Yandex told us.
The first year brought us 10,000 visitors, or several dozen visitors per day.
Second year – 156,000 visitors. There were several hundred visitors per day on average, with peaks of up to 1,800 per day at the end of the spring session.
Third year – 391,000 visitors. It is worth mentioning that the beginning of the year was very successful, there was an almost linear growth of traffic and almost 4,000 users per day at the end of December 2023, but everything ended after the New Year 2024 and the Google search engine update. Traffic from there fell 3 times and returned to the level of the second year.
Traffic in the third year of the project's life This was a serious blow, because we were already thinking about how we would approach the desired level of unique users per day by the beginning of summer, implement a mechanism for targeted advertising with high conversion and start making money.
Why did this happen?
It's hard to say for sure. My best theory right now is that it was influenced by the fact that we have a relatively large amount of unoriginal content on our site. We tried to fix this situation, rewrote some content, took measurements, and evaluated the results. We managed to improve the situation a little at some point, but it's far from what we expected.
It is also worth mentioning that throughout the life of the project we have been engaged in technical SEO optimization of the site. At some point, we managed to bring the technical performance of the site's content pages to the level of Wikipedia, so that search engines would like it as much as possible. We have implemented many different things for SEO according to the best recommendations of Google and experienced SEO specialists, whom we attracted for consultations.
Content Page Performance Metrics Overall, it was very unexpected for me. I was sure that having good SEO performance of the site and user-friendly interface we are doomed to success in search engines. It turned out that this is not enough.
Perhaps we did not pay enough attention to external promotion, but relied heavily on organic growth.
What would I do differently now?
Before starting a project that relies on high traffic, I would find out how it is planned to achieve this. I would set aside a much larger marketing budget if possible, I would think about an external promotion strategy. I would also involve experienced SEO specialists from the very beginning so that they, together with the designer, could help in building the correct structure of the site, so that I would not have to redo some things after the project has been launched. I would definitely not rely on competitors' figures without understanding how and at what cost they achieved them.
And perhaps I would question the implementation of such a project at all, which can only be successful if search engines like it. Search engines cannot be relied upon, no one ultimately understands how they work and you cannot build a business model on this.
Honestly, it feels like the way search engines work is an evolving black box and it's like no one really understands what needs to be done to make them like it.
Lack of methodology for the development process during work on the first versions of the project.
There was no task estimation, no planning. We only had a rough idea of what we were doing in the next period of time, and often it was not limited at all. We kept some cards in Jira, but nothing more.
This led to us constantly delaying deadlines, there was always a feeling that we were not on time, this disappointed my partner and me too. And one day there came a moment when it became clear that this could not continue any longer – we could not be relied upon. At that time, there were 3 people in our development team.
This was the turning point for me and our team. I realized that we needed to manage the development process properly, and I didn’t have this understanding before. I lived in the position of a developer and didn’t ask myself such questions.
I took many courses on Scrum, understood it thoroughly and understood why it is needed and why this approach is so popular. I understood how it can be used to solve the problems that we have. I understood why they arise. This allowed me to independently implement full-fledged Scrum in our development process. After that, I needed time to train the team, explain why we need these ceremonies and set the bar.
Since October 2022, we have been working using Scrum. We have made huge progress in terms of the quality of backlog reviews, sprint planning sessions, our estimates. Since then, we have never failed anyone's expectations, and have always delivered what the whole team signed up for.
We were passionate about our vision for the platform, not the business outcome.
And, of course, we were very interested. However, initially we wanted to develop a commercial project, but it turned out to be social and charitable.
I think this happened because we relied little on feedback from reality, often made decisions based on our vision and thought little about the commercial component of the project.
What would I do differently now?
If the goal is to make a commercial project, then I would first of all concentrate on this. On the business plan and on the product features that will allow it to be implemented. I would try to direct the project towards its development being determined by the needs of users. I would build development processes around this, transmitting this understanding to the team.
Conclusion
I am convinced that it is necessary to share not only successful stories. Because you can't draw any conclusions from a successful story. But from one like mine – a wagon and a small cart.
I hope that my article was useful and helped you. Maybe it gave you some ideas.
In the future, I plan to talk about the technical side of this platform, as well as about building a team.
If you are interested in anything or have any questions, I will be glad to answer in the comments. And if you want to talk in more detail, here is my LinkedIn And Telegram.
Thank you!
