5 mentor's tips for launching a startup

Whilst being a mentor at the business incubator Eastlabs, Daniel Tonkopiy repeatedly gave useful advice to young businessmen — startupers. He mentored, Preply, Poptop FM, and at the same time was launching his own startups — virtual games X-Rift, video search, online guestbook, and finally, his most successful project — courier delivery service Delfast. Daniel has got some messages and advice that will come in handy for future businessmen, this content was especially made for Marketplace Wiki by Rademade.

1. Solve the problem you have personally faced

The first online project that I worked on, which I founded, started its life in Kyiv. The website gathered information on city restaurants so that the users could rate them — by location, expensiveness, style of cuisine, and service.

I had such an idea because I often traveled from Kazakhstan (my hometown) to Kyiv and could not find a suitable place to eat lunch. So my personal problem became a basis for the business.

About 100K users visited the website each month, and the number was increasing 1-2% per month. However, these were not actually the figures we were expecting, therefore we decided to sell the platform. The project was bought but within a year it was closed down. The situation was similar to when Google acquired the Ukrainian project Viewdle for $45M and unexpectedly shut it down.

Nevertheless, I launched the next startups solving not imaginary problems but those dictated by my own personal experience. This is the only way to create something viable, as only in this case are you an expert on the matter and can rely on yourself for innovation.

2. Try to predict — count the amount of piano-tuners in Chicago

There is a useful exercise which helps you to calculate the number of piano-tuners in Chicago (according to Fermi). I'm suggesting that we do the same task for Kyiv in order to understand how to predict the market coverage.

So, you have to make an assumption because you have no input data to calculate with. Let's take the population of Kyiv of 4 million of people as a basis and assume that every 5th family owns a piano.

An average family consists of 2.5 people. We take 4 million, divide them by 2.5 and get a figure of 1.6 million families in the city.

Now let's calculate the number of pianos in the capital. According to our assumption every 5th family has this instrument. We divide 1.6 million into 5 and receive a figure of 320 thousand pianos in Kyiv.

A piano-tuner fixes 1 piano for around 3 hours. During the day he works on a maximum of 3 instruments. That adds up to 15 pianos in one week, 60 pianos in one month, and about 600 per year (taking into account his vacation).

Let's divide 320K (the number of pianos in Kyiv) into 600 (instruments that can be fixed by the piano-tuner for a year). That makes 530 piano-tuners.

Each parameter has assumptions and the more inaccurate the statistics are, the higher the deviation percentage. However, with this calculation we can make an assumption concerning the market. It's better than nothing.

3. Test your idea on users

There is no other way to test hypotheses but to ask end-users.

As soon as the idea of the Delfast startup arose, I went to the clients. I had a hypothesis about a quick delivery of parcels with electric bicycles. The only thing that we needed to know is whether anybody needs it. I asked the CEO’s of 10 Internet shops (,,, etc.) what they do to improve their logistics. I offered them delivery for 1 hour at $3 — 7 out of 10 clients decided to buy such a solution.

Pay attention that when I communicated with representatives of trade networks, I had neither electric bicycles, a team for the project or a business model. Only the idea and preliminary estimations. But I tested the hypothesis from the start, that's why the project is successful not only in Kyiv, but also in Almaty and Warsaw.

4. Read negative feedback about competitors and suggest UVP

Feedback about your competitors should become food for thought. If users write comments that the watch is great but the strap is broken, you should produce the same watch but of higher quality.

For Delfast delivery service, the feedback about competitors played the same role. The clients complained of delays when using the express delivery service, therefore we decided to become faster. As soon as we managed to deliver parcels within 1 hour, it became our unique value proposition — “We deliver as soon as pizza delivery”.

A project should have a UVP, otherwise you won't win on the market. For example, the DuckDuckGo project could compete with Google because it found its UVP — a possibility of the anonymous search.

If a project has no uniqueness, you should not launch it onto the market. For example, there was no sense in introducing the marketplace, which launched in Kazakhstan, in Ukraine and Russia. There were already websites for industrial purchases and we could not offer anything more.

5. Don't start the project solo

990 successful startups out of 1000 were launched non-solo. The project should always be run along with a co-founder and you should choose loyal people to build the core of the team. From my own experience, the quality of the team is the first thing considered by mentors and investors. When I worked at EastLabs and cooperated with other incubators, I was thoroughly learning how people worked. I worried if this team will successfully manage to realize the idea as it was originally intended.

That's why before asking for investments for the project, you should build a team, develop competitive advantages of your business and find out whether the market is ready to buy your idea.

And one more piece of advice — place yourself in the investor's position. Assess the situation from his point of view and think if you would invest your savings into the project. There is no sense in applying for help unless you are ready to take a risk. To launch a startup, you should risk it all.

Header photo: kritiya /

Related Articles

Recommended by MP Wiki

Do you want to launch your own markeplace and need some help?