heureka!grouph!g

Data Analysis

Data is a huge help, but it’s not all-knowing, says Heureka’s data chief

25. 06. 2021 | Adéla Berková

Heureka started on our market as a price comparison site, but it is increasingly presenting itself primarily as a shopping advisor which operates in a broader field. It is now active in nine markets in Central and Eastern Europe, making it the largest price comparison site in Europe, and it is also one of the top three players in most countries in this region. Thanks to this scale of operations, Heureka as a group has a huge amount of data at its disposal, which it now wants to offer more to third parties in aggregated form.

„We have interesting data samples on what’s happening in the market in each category,” said Michal Buzek in an interview for CzechCrunch. He joined Heureka less than four years ago, and as the Chief Data Officer, he is now pursuing a strategy to offer insights about what is happening on the market so that e‑shops, brands and other companies can make better decisions. For a similar reason, Heureka recently purchased the Brno analytical start-up Dataweps.

„The market sees that we have a lot of data, whether from monitoring conversion rates or from the ’customer verified’ services. Now that everyone knows this about us, and some people have also been questioning what we’re doing with the data, we decided to aggregate it at the product and category levels and start giving it back to the market. So that e‑tailers also have the opportunity to map the situation on the market, they can connect the market data with their own data and from this obtain interesting information,“ said Buzek.


In the interview for CzechCrunch, the chief data officer for Heureka explained how e‑shops and brands work with data in general, what they should focus on, when data is not all-knowing, and also in which data areas Heureka feels it is strong, and conversely, in those which it feels weak.

Why has Heureka now decided to actively promote the possibility for third parties to use your aggregated data?

We wanted to show the market that we’re not just the bad guys who only collect data, but that we’re able to give it back and provide players, whether they’re e‑tailers or brands, interesting insights into how the market works and what categories or products are growing. This was one of the strategies and goals that Tomáš Braverman (CEO of Heureka) and I put together when I joined Heureka.

Does aggregated data mean you don’t sell information about individual users and e‑shops?

Yes, exactly. We would never risk leaking e‑shop data to other entities. We aggregate them at the product and brand levels, and ultimately everyone benefits. Of course, we try to leverage our advertising space with user data, but we don’t trade or share this data with any third party.

Let’s take a closer look at how e‑shops actually work with data. What areas do they most often concern?

Data is most often dealt with in marketing, and monitoring of competitors’ prices is definitely a strong area. In recent years, many tools have been developed to monitor prices and notify e‑shops of any changes. Related to this is the trend towards dynamic pricing, which is where automated tools can change prices themselves according to set rules.

Larger e‑tailers who have sufficient capacity or data-driven approaches are increasingly looking at management of the product mix – regardless of whether they have the right products actually in demand. They usually know what products customers want directly from them, but when they look at the market data, they may see other products in categories they don’t have in their portfolio at all.

„In general, I see a big future in the automation of e‑commerce, and our acquisition of Dataweps is related to that.“

Michal Buzek, Chief Data Officer Heureka Group

And you can provide this exact data to them?

Yes, we can show them what products are missing or what products they have which there is no demand for because they don’t promote them in their advertising channels and are therefore not visible. Sometimes you just need to focus on the right products to get the given e‑shop to rank higher with us or on Google.

Marketing has been „data-driven“ in e‑commerce for a long time. Is it correct that there are many tools on the market which address this area?

That’s exactly right, automated bidding tools in marketing channels have been around for years. At the very least, companies are able to do this efficiently on goods comparison sites and at prices that are affordable even for smaller e‑tailers. For example, they may decide not to bid for the top listing positions with their entire portfolio, but only with the 50 highest margin products. Many B2B partners are developing tools for this type of automation.

Are price monitoring tools like this already widely available?

This is more a matter of the medium and larger e‑shops and especially brands. They are able to check how individual sellers offer their products. In general, I see a big future in the automation of e‑commerce, and our acquisition of Dataweps is related to that.

Does this type of automation also apply to price setting, i. e., dynamic repricing?

This is a trend which was started by the bigger players such as Alza, Mall, and maybe also Datart. In the last year or two, several tools have emerged on the market, and companies that have been monitoring competitors’ prices are now only a small step away from dynamic pricing. They’re now raising awareness in the market and trying to reach as many customers as possible. The trend has definitely already started, and now it’s moving from the big players to others.

In this respect how does Czech e‑commerce stand in comparison with other countries? Are e‑shops here considered to be data advanced?

I agree that Czech e‑shops are quite far ahead in terms of the countries east of us. For example, in Hungary, the situation is quite different. E‑shops there are just now getting to know what automated auctions are, and the topic of dynamic repricing will come up in maybe two years’ time. Although this country is not far away from us, their situation with e‑commerce is definitely different.

If e‑shops here are accustomed to a certain standard and functionality in systems, could it be a problem for them to expand to Hungary, for example, because the tools there are not so advanced?

E‑shops used, for example, for automated bidding (bidding in auctions for the best search positions – editor’s note), could move into Hungary with the idea that they want to do the same, but the style of auctions is slightly different there. Dataweps is already focusing on the Hungarian and Romanian markets, which is why we acquired the company in the first place. We have big plans to accelerate automated bidding in these countries, and we want to teach this to local e‑shops as well.


A wealth of tools and data is available on the market. How should an e‑tailer calculate whether it is worth the investment and how advanced should the tool be?

When it comes to integration of marketing data, there are plenty of digital agencies and freelancers in the Czech Republic who can help e‑tailers. For a few thousand or tens of thousands of crowns, they can set up reporting for an e‑shop. The question is whether the e‑shop is really ready to work with the given data.

How can that be determined?

When e‑shops are small, they solve problems which are completely different from working with data. They need to survive, ensure distribution and the logistics processes and warehousing. Then they reach a stage where they automate and fine-tune their internal processes, and with further growth, they can improve their performance and expand their portfolio. For that, they then need data. We see e‑shops trying to work with data where there is a lot of competition, for example when they sell mobile phones, tyres or pet food. In these segments, we have the greatest demand for data.

What other factors affect the ability of e‑shops to work with data?

Of course, the owner or manager of the e‑shop affects this. It is much easier to work with data if they are so inclined, and then they will find people to do the work. It is very important to have skilled staff who can make data-driven decisions. If they do not have such staff, then it makes no sense to invest in data solutions.

However, many people think that they really know how to work with data, even though this is not actually true. They arrive at a conclusion according to their opinions, but they may not be correct. How can this be avoided?

When we show data products, we consult with our clients and show them what to focus on – the types of scenarios, how to use the reports. Otherwise, I would really recommend arranging several hours of consultation with an agency. Unless the e‑tailer is data-savvy, it’s always a good idea to get someone else in to do it at the start. After learning to look at a few basic things, it then helps in decision making, company direction and business thinking. I always enjoy showing people the power of data and what can be mined or gleaned from it.

What are the most important indicators that e‑shops should monitor?

In the majority of cases, the most important indicator is the costs to sales ratio, so that the e‑shop knows whether it is profitable in a given marketing channel. It can also be calculated in a sophisticated way by factoring in the cost of the specialist who helps with it, or just simply based on the marketing costs it sends to the tool. Usually e‑tailers look at the costs to sales ratio of the entire channel, but then it is definitely important to also look at this metric at the individual product level and compare it to the margin. When I see that the costs to sales ratio is 20% but the margin is 15%, then I know that I can never be profitable and that I should focus on those products, perhaps change the bidding strategy or focus more on conversion of products.

In the context of product data, it occurred to me – if you get ten e‑shops from the same category and advise them all that five particular products are selling well right now, won’t they all start selling the same thing?

At Heureka, we don’t advise what to sell, we just provide the data, and it’s up to the e‑tailer to choose their tactics. Generally speaking, on comparison sites, e‑shops can define themselves according to product or price. In these cases, they’re targeting the lowest possible price and have to watch their margin to make sure they aren’t losing money, or they’re holding a higher position and trying to stay in that prominent position. It’s always about the specific strategy of the given e‑shop.

„Smarter e‑shops no longer just want to have the lowest price.“

Michal Buzek, Chief Data Officer Heureka Group

Do you see any trends in fewer e‑shops trying to compete on price and focusing more on other areas?

Smarter e‑shops no longer just want to have the lowest price. Dynamic pricing tools should also advise on when not to lower the price and still have good performance and profitability. Losing a few orders that have smaller margins might be better sometimes, but a lot of e‑tailers don’t realize this and think that if they keep the price low, then that’s the way to succeed.

Recently the trend has been to balance the price reductions so that it is still worthwhile to do so. Data can generally help speed up and streamline internal processes, and in managing the product mix, it can show a number of interesting things. It’s up to the staff of the e‑shop though to decide whether a product they don’t have in their portfolio is in line with their concept.

Do you have an example?

We have a large pet food retailer who bought a report from us about missing products, and they pointed out to us that from a total of ten brands, they did not want five of them in their portfolio, because they were, for example, „low cost“ or made from lower quality ingredients. There has to be a human decision-making process which determines what is and isn’t suitable. For example, if an e‑shop is losing on a product, it may use it to gain market share and prefer to skip profitability.

You mean combining data from several sources?

These days, this is the big strength of data. If I analyse Heureka’s performance without market data, then I will only get a narrow angle of view. When I add in the market data and external influences such as closed high street stores or weather, it gives me much better context. That’s what we try to do at Heureka with market data, we don’t just provide reports but raw data so that e‑shops can integrate it into their systems. Heureka is moving forward in this aspect, and we want to keep doing so.

When we talk about sources of data, there is always a question about the quality of data. Even the best algorithms and reports either thrive or fall on the data we feed them. How can e‑tailers know what data to trust, and how can they monitor its quality?

When we negotiate the buying and selling of data, we provide clients, for example, with test data for a historical month so that they can be confident it makes sense to them. Occasionally, an error occurs, and we try to clean it up with the client, for example, when a product is split into several cards. We try to make sure that we both trust the data. Our aim is not to sell data to the greatest number of e‑tailers possible and then be held to account because the data is deemed useless.

Heureka may have an apparent data monopoly in this area to some extent, and if you make a mistake, then it can harm the e‑shops. How can you protect against something going wrong?

We lay our cards on the table, we show the methodology, we show how the data is generated, how many e‑shops are involved in the category, and what proportion we have to calculate because we don’t have exact data for all e‑shops. Heureka is, of course, a strong player generally. We cover about 2,500 product categories, but for 700 of these, we don’t even try to offer data, because we don’t feel that we are so strong in those categories.

What segments is Heureka strong and not so strong in?

Heureka and Czech e‑commerce generally is strong in electronics. We are very strong in the home and garden segment, while sports and children’s products have seen tremendous growth. We are weaker in food and drinks – these are daily consumer goods, and we will probably never be strong there. On the other hand, we are quite strong in alcohol, because when someone want to buy a bottle of something good, they try to find where it can be bought for a cheaper price. Building materials is an interesting segment. It is complicated because there are many unique products and there is not much competition. Someone might order 200 tiles in one order, someone sells them by the metre, so it is difficult for us to collect this data. We therefore don’t go into this category.

You haven’t mentioned one of the biggest segments – fashion. Where does Heureka stand in relation to fashion?

In the last two years, we’ve focused more on the fashion segment and also the furniture segment. Historically, we haven’t been very strong in these areas because they weren’t attractive to users. Now, we can better sell photos, and we’ve started working with a company that recognizes clothing parameters based on pictures. Describing the colour, sleeves, cut and so on. This has helped us improve our data, and we’re planning several activities because fashion is one of the segments where we want to become stronger. A lot of people think that fashion is just Zalando, About You and Zoot, but many customers buy clothes from Lidl, Sportisimo and other similar shops. We want to educate people to use Heureka when they buy in these categories.

In your opinion what is the future of data in e‑commerce?

We demonstrated this with the acquisition of Dataweps. We believe not only in automating the marketing process but matching the products to catalogues or databases. Dynamic repricing is and definitely will continue to be a big theme in the coming years. The questions are always, what are the biggest players on the market addressing and what is the rest of the market addressing?

“Big international brands have started orienting themselves more towards e‑commerce.“

Michal Buzek, Chief Data Officer Heureka Group


How do they differ?

Automation of basic marketing processes, management of the product mix and possibly also pricing will be the big topics for the next five years for smaller players, while the big players will be tackling more sophisticated problems, such as personalization, offering alternatives and different products, how to raise the value of goods purchased in a single order. There are already platforms for recommendation and personalization, but I think that working with the customer is just beginning and is pretty bad for many e‑shops. I see this as a great direction, which is why everyone is trying to make marketplaces, to know as much as possible about their customers and to offer them the best possible options for shopping, services and other things they might be interested in.

We’ve talked a lot about e‑commerce, but you’re also interested in market data for the brands themselves. What does it actually address?

The last „Covid“ year was interesting. Big international brands have started orienting themselves more towards e‑commerce. The big high street stores they were previously focused have closed, and these companies have suddenly found that the baby milk or pet food segments are very important in the online marketplace. They’re now trying to recruit specialists and strengthen their e‑commerce departments.

So previously they weren’t so interested in e‑commerce?

They were mainly interested in offline sales, and suddenly they’ve started contacting us, asking what the situation in e‑commerce is, or that they have a bad market position in a given category. At the same time, we also have data on customer experiences with products. This data can tell them that they are failing because some of their products have below-average quality ratings.

Is this also a long-term trend?

E-commerce is not just about e‑shops. Even big brands have started to understand that a large portion of sales is happening online and will continue to do so. This is the change that has accelerated digitalisation at large companies. And that can only be a good thing.

Original interview published: https://www.czechcrunch.cz/2021/06/data-jsou-obrovsky-pomocnik-ale-nejsou-vsespasna-rika-datovy-sef-heureky-chytrejsi-e‑shopy-uz-nehraji-jen-na-cenu/

Share