How To Make Powerful Market Analytics Based On Insights And Open Data

How To Make Powerful Market Analytics Based On Insights And Open Data

I was doing research and strategy at Skyeng, and now I am developing my business and helping various companies to make data-driven decisions.

One of our favorite clients is Yandex.Praktikum. In August 2021, we did a small but beautiful study for them. I want to tell you about it.

What Is The Task

The WorkshopWorkshop is a large and well-known educational service that is actively growing. But it is better to grow in the directions with the most money (considering the direction of work, of course). What courses should you launch first to recoup them and increase revenue faster? It’s helpful to look at which courses make the most money for your competitors.

The WorkshopWorkshop contacted us to do a market analysis. It was important to understand the volume of revenue in the context of each profession, first focusing on large online players. It is clear that there are no such figures in the public domain – but there is a hypothesis that we can get and test them.

Training

First, we looked in detail at all competitors and the specifics of online education.

The Workshop, Skillbox, Netology, and other players usually have courses, but professions. Courses are inexpensive and designed to build specific skills (for example, a service editing course for copywriters). And professions are complex, expensive, and long, help to get a profession from scratch on a turnkey basis, often overlap with courses and consist of them (for example, “Become a data scientist in 1.5 years”). These are different products with different revenue scales.

After talking with market representatives, we realized that the main income for companies comes from professions. Therefore, we decided to focus on them.

Next, we parsed all the courses from the companies’ websites. To compare the professions of different players with each other, we even compiled a single dictionary. Because the same professions can be called differently in different services, Out of about 500 titles, we came to 94, which overlap in the top 10 players we want to analyze.

First Iteration – Attribution

Our task is to understand how much money each profession generates for all major market players. Roughly speaking, how much the market makes from training Python specialists and data scientists.

But to understand this, you need to understand the revenue for each profession in each company. The total revenue of the players is mainly in the public domain (and where it is not, we found out our ways). To correctly attribute it to professions, we wrote a model in Python.

The model receives useful factors; it tries to predict revenue by direction at the output. Among the factors we considered were:

  • Searches for Words that. The more the demand for training in Internet marketing among potential students, the more requests on the network
  • Several courses. If a company launches several courses with similar names, we consider using a correction factor.
  • Price of courses.

And other parameters too. As a result, our mathematical model has identified the shares that certain professions occupy in the company’s revenue. But we did not understand yet how accurately she does it and decided to check the results with the help of insiders.

Second Iteration – Insider

The best way to understand how much your competitors are making is to ask your competitors. And competitor’s rivals. This is a fairly simple and effective method, which, however, has its subtleties.

We love to collect insights and do it well. It is important to collect information honestly, openly, and mutually beneficial. Like this – don’t:

– Hey. Lesh, we are preparing one study here. Tell me, what is your income by profession “Data Scientist”?

– Misha, are you sick? Could you not write to me again?

Better to go to Vita first:

– Vitya, hello. What do you think is Lyosha’s LTV by profession “Data Scientist”?

– I don’t know, I suppose that in the region of 100-120 thousand rubles.

And then you can go to Lesha:

– Gosh, listen, do I understand correctly that LTV by profession “Data Science” is around 100 thousand rubles? Is it true?

– Hard to say.

– Well, do you have the same order, or much more?

– Yes, it seems.

– More or less?

– Well, why are you sticking? More, a little more. I would like even more, but it’s ok

All these, of course, are exaggerated examples, but they show a general approach (in fact, we communicate, of course, not so intrusively and ask questions more carefully).

By scrupulously communicating with people, you can collect a lot of useful data right in the numbers: the company’s revenue, revenue in various areas, in our case – for individual courses and professions. This is not rocket science; anyone can do it; you need to talk and be polite.

We know how to do it well because we know people in the market and, in general, have got our hands on it: we know how to be persistent and accurate.

But then there was a problem. The results of our model did not fight with insiders. At all. The fact is that the model relied more on external indicators of demand, and it lacked “internal” business indicators. Historically, different companies have focused on different areas (for example, Geekbrains – on programming, Skillbox – on design and marketing, and so on). The model sees all the professions on the site and does not understand the flagships.

We could manually correct the point values ​​based on the received insights and even retrain the model on them – but this would be a one-time crutch solution, and we wanted to get a tool for regular updates.

Therefore, we sat down to think about what publicly available indicators to take into account the company’s internal focus on certain professions.

Third Iteration – Advertising

We brainstormed and realized where to get the exact data – online advertising. We climbed into the Facebook Advertising Cabinet and realized that advertising traffic could be directly related to the “money” in courses: they advertise what brings the most revenue. We started to dig in this direction.

However, there are no numerical indicators on Facebook, so in the end, they took search traffic from Google and Yandex.

Parsed course pages and linked them to traffic. We used advertising traffic, organic search traffic, and the course price for the model.

We are lucky that usually, each course lives on its page, which means that we can smoothen traffic for a specific keyword with a page URL from site scraping. What we did – by retraining the model on the new data.

The model yielded new data that correlated perfectly with both the Workshop data and the insights of other players. Bingo!

Additional Iteration – Enrichment

We decided to improve the model by enriching it with data . We parsed the skills that are indicated in the description of the professions. As a result, they showed in what directions to work and what specific skills it is important to train in each course. This is an important clue for the methodologists of the WorkshopWorkshop.

We have been working on this for just over a month. We moved forward iteratively and collected everything in a large report and updated tables. The final scoring considered the market, potential, launch complexity, and brand fit.

The colleagues from the WorkshopWorkshop were very pleased – our results will have a great impact on their expansion in 2022. We continue to work in several new directions at once, in Russia and abroad.

Instead Of Totals

A similar task can be done by any self-respecting marketing director (himself or with our help). I will give a couple of tips:

  • Develop your network, meet competitors: in social networks, at conferences, in person.
  • Share information. It is not necessary to take something out of the NDA; you and them do not have anything to stop you from commenting on the news and your data, sharing the news. We are not only competitors but also colleagues in the market. We are moving the entire market forward, and for this, it is important to share information.
  • Always test multiple hypotheses. Make three approaches to the machine, reject data that give too much deviation.
  • Build your team’s marketing research skills. Hire staff researchers and outsource consultants.
  • Make more data-driven decisions. The famous guys in the market live on data – it’s useful to learn this.

Also Read: Why Advertising Is Needed

Pure Tech info

Pure Tech Info is a Unique Platform that regularly keeps you updated about the latest technology trends, business awareness, product reviews. Also, information related to the latest Gadgets, App's, Cyber Security updates, latest Digital marketing tips, Marketing Ideas, Tech news, and many more categories. It's a website that provides the best and pure technical content to the readers.

Leave a Reply

Your email address will not be published.