Roman Nester, co-founder of Segmento - about the future of advertising technologies on the Internet

Roman Nester has a degree in social psychology of mass communications. From 2004 to 2009, he worked in the display advertising market in companies of foreign holdings WPP and Bonnier. Since 2009 - co-founder of the company-developer of trading financial algorithms QuantBrothers.

In 2011, he became a co-founder of the Segmento company (in 2011-2014, also worked under the Rutarget brand), the controlling stake of which was acquired by Sberbank last spring. Roman is an expert in predictive marketing and programmatic strategies.

According to the widespread point of view, RTB is such a “deaf forest”: very important things are happening in it, now some kind of fabulous creatures start to run there, which will look like “unicorns”, as start-upers say. But still there are more gray spots than answers.
')
We decided to address the questions to Roman.

- What is RTB, in your opinion?

The question is good ... Let's just say: I believe that we are not engaged in RTB, but in the advertising business and advertising technologies. Because spinning around the abbreviation RTB is just the same as saying: “What do you think about HTML? Is HTML promising or not? ” In fact, HTML is just a language in which you can create completely different websites, solve different problems, and usually few people discuss it separately. At the same time there are huge production agencies, sites that have won something, which are made on this HTML.

Similarly, RTB. RTB is just a protocol that allows you to conduct an auction for an advertising display in real time. There is nothing radically breakthrough or new in these technologies themselves, but the emergence of such a protocol and its mass distribution (the transition of almost all major mass networks in the world to, let's say, an open ecosystem, when a huge number of automated networks fight for the demonstration of each user) gave a new step advertising, a step that allowed to go to the level of an individual user.

If earlier we could buy packages of sites or any one separate site or try to buy a whole set of automobile sites, now we can buy a show for an individual. And this approach itself has greatly changed all the opportunities that big businesses have that spend money on attracting users and advertising in principle.

- The first point that I see here and which needs clarification is the following: well, an individual approach was convinced. But how does this ultimately differ (we are now talking specifically about your company) from what is offered to me by fairly large advertising platforms like Google AdSense?

After all, in an amicable way, they also sell some target audience. They say that it is just as clear to them through the use of systems that analyze user actions, cookies, their search queries and the rest of the data. What is the immediate advantage for an absolutely average, impersonal client?

The fact is that different customers need very different products. Speaking of what AdSense offers to the audience ... AdSense is, firstly, those to whom we have to say thank you for spreading the open auction approach and the rise of RTB technologies, because AdSense taught Russian sites to automatically sell their shows. Then Yandex pulled up to it a little later, and this allowed us to create the largest advertising network of shows in Russia.

If earlier you had to recruit platforms, place some codes, go into ten different ad networks, then it turned into one giant network: Google AdX / Adsense + Yandex cover most of Russia.

Thus, we, as an automated platform for the purchase of targeted advertising, were able to make up to 20 impressions per day for the user that we need. Before such a possibility was not. This is the first.

Then, if we talk about the products of Google or Yandex, these are companies that make one clear and simple, like a stick, product. In hundreds of countries, it must be standardized in order to work correctly everywhere. In addition, this product was originally aimed at small and medium businesses.
If you recall, in The Internship, about Google, in the end, these interns are asked to sell the AdWords service to a small pizzeria.

And we are a company that aims to provide marketing services and services for the segment of the largest advertisers in Russia. Our tasks are different from the tasks of Google, which work on incoming demand, our work is the creation of expert services and solutions. RTB, as I said, is just the foundation of the solution. Being engaged in this for four years, we realized that we, better than anyone else, are able to solve the problems of large advertisers using such a set of technologies. And we focus on that.

What makes us different from Google, besides the approach to customers with whom we work?

This technology and data. We see our main advantage in taking a step further than just marketing online than simply showing ads to people who have visited some pages. We are sure that the future is not at work with cookies, but at work at the level of real people. We are already doing projects, combining both worlds: online and offline, which previously lived completely separate. This is a huge, time-consuming task. We know that all the largest companies are trying to move in its direction, but we still believe that a fast and flexible team will do it better. I mean, we bring the knowledge of a huge number of people offline in online marketing. Online marketing has not yet fully realized what is happening.

- Here, for example, you have a box. This box has a set of tools. You get in touch with the largest advertisers and tell them that with the help of these tools, you can most effectively solve those marketing tasks that they have already solved.

At the same time, you are saying that you use better data about your audience: online + offline and the rest of the array that you have. Will you prove it?

We rather tell them: “Before, digital was an auxiliary tool for you, and you didn’t trust it, and in fact - 90% of large businesses, despite the fact that they are trying to be digital-oriented, they still don’t trust the Internet.” Therefore, we say: "You used to work with cookies, with websites and shows, now we offer to work with real people." The fact that a person is real is our business strategy. Our strategic partner is the largest bank in the country. We are in a unique position: you will not find such a penetration of a dominant bank in any country.

This bank allows us to give the advertiser a guarantee that he is contacting a living person through his digital channel. Well, to further develop this story to the point that before the desire to use the level of income, the level of spending or to predict the consumer's intention to spend money somewhere was not available anywhere else. We do not just come and say that we have some special data here, this is in principle a different approach to digital procurement.

“How do you work with people then?” Can you tell about it at all? Or is it a trade secret? How does this interaction take place with the final person, who is also a consumer?

It is important to understand that our company does not have direct access to personal data or information protected by law on banking secrecy. Working with primary data takes place inside Sberbank, we get ready-made segments anonymized by the bank, grouped according to a particular criterion. And it is with these segments that we continue to work. Regarding the details of how we combine the profiles of people from the segment formed by the bank with their online profiles, I can say that for almost any bank, identifying the profiles of their clients on the Internet is one of the key tasks today. Enormous money is invested in this and the work takes place inside the bank’s security loop. Our solution, implemented by Sberbank, allows us to preserve the efficiency of data analysis, but at the same time withstands the most serious requirements of the bank’s internal services. Also an important point - Sberbank is our shareholder and has a controlling stake, it allows him to see and control all our internal data processing processes.

- Do I understand correctly that you are receiving transactional data that has been anonymized in some way?

It can be said that we obtain derivatives from transactional data for targeting on the Internet, because it is impossible to transfer transactional data directly.
In addition, having the opportunity to see the full behavioral path of each user online, we know everything about him that Google knows, Yandex or Mail.ru - everything he does on the Internet. As a result, we internally connect the history of offline and online behavior.
Comparing these two things (offline and online) gives a different level of work with the audience than was available before. This is no longer just targeting and trying to get some clicks. These data make it possible to understand as a result of a large advertising campaign whether people have made new transactions or not, that is, to measure the direct economic effect of advertising.

Inside we call it the “closure of the marketing ring.” Usually in one place marketing spends money, in another place a business sells, and they are connected in each part, but not closed: as a result, the business does not know exactly what it received from marketing spending, and marketing does not know exactly what it is from his efforts went on sale. We are the missing element that this cycle closes.

We sometimes joke that after we finally realize this (we will get access to the composition of the check in order to understand what specific products a person has bought), there will be nothing left to think up in marketing.

I look at what the giants are doing (Google, Yandex, Mail.ru, Criteo abroad): everyone is trying to go offline and see if they visited the mall, if the people to whom they showed advertisements bought something.

- Advertising campaigns are very different. When you talk about marketing, I hear about what you say about standard ways. I imagine it this way: I saw a billboard somewhere, and after that you try to understand if I really saw it and how it affected me in general. Do you somehow predict customers' ROI and themselves to what extent do you believe in it?



There is no need to believe. In fact, we are also speaking with public cases on this topic. From the first day it was important for ourselves to understand whether our work brings the result to the advertiser.

After all, as it was before: display advertising has always been an unloved child of business. There was a TV in which you could invest and believe that it gives a correlation with sales. There was an incomprehensible Internet, in which search advertising appeared at some point, which gave advertising a direct response.

“Mediyka” and video (and this is a huge amount of money and a huge number of impressions) all the time were hanging around the side, and no one wanted to approach them to include them in the overall performance picture, in order to understand what they really bring to us.

We can be proud that, to date, in almost all areas of our clients' business, we have received meaningful and reliable metrics of what media advertising brings to sales. If you take a look, Google now comes around the world with educational presentations telling how important big companies are to invest in media. In order to search (the main business of Google) at least one of the potential customers asked about the company and its services, the question must first be resolved to ask for some reason. And display advertising is what becomes the first point of contact.

I did not just say that our focus is the largest advertisers. This is also due to the fact that our sale is an expert: you can’t just come and say: “This is the product, press the button and sales have gone there”. In order to properly do our work, we really need to understand the entire chain of contacts of this business with its consumer, figure out how he is investing in the media, tell him what tasks can now be solved using a programmatic approach and how this will affect the result: how many people will come to the site, how many of them will convert, how many will convert offline.

This cannot be done using the standard, template method. For every business, this is a new job every time. But we have received such an answer for each segment. And I have not found a single business segment where banner and video advertising would not have resulted in the resultant effect on other channels: search, SEO - on everything that people usually buy on the Internet, and then offline.

That is, our work is difficult. It is not only about efficiently targeting and getting more ROI on media investments. It is also about how to learn together with the client and teach him how to build the attribution of what is happening to him - which channels actually influenced his sales and how. This picture is now very few people see in reality. Attempting to find the most effective channels is a very strong simplification of the picture of the world. Therefore, we have to not only offer our product, but also engage in educational work.



- How long does your company exist as an idea and enterprise? How long have you been doing this? What is the size of the team? How did it increase? What happens to you after the transaction with Sberbank? How well do you scale?

In general, the company can be considered the year 2006, when we and partners (with whom we are now doing Segmento) launched the first story of trading algorithms in the field of high frequency trading. It was something similar to what is happening now in programmatic: these are algorithms that work in the derivatives market, which take decisions in a split second: buy, sell. If so, at what rates? We have been actively growing this business for five years. He exists now.

In 2011, we saw that a similar trend was taking place in the advertising market: automation appeared, programmers, like us, were able to enter a new market. Likewise, it has prevailed with the stock and derivatives markets. By the way, in the derivatives market, as a result, up to 80% of operations began to be performed by robots. I think a similar story will happen to the advertising market soon enough. Just as people lost their jobs in the traditional brokerage market, they would lose it in the advertising market.

In 2011, with this background on the development of algorithms, we began the development of a new technology, tied to the collection of user network data, behavioral data processing, and advertising targeting. Initially, we seriously aimed at creating a very high barrier to entry. We spent a lot of money on data and development expertise. Although we could invest them as many competitors in marketing or sales. But it bore fruit.

We are a company from Petersburg. Only in 2013, we fully moved to Moscow. And so far we are growing and achieving our goals thanks to the St. Petersburg team - these are our developers, data mining specialists, and the guys who are involved in the implementation of advertising campaigns. I am convinced that this team is one of the strongest in the country in terms of experience and a combination of professional skills. This division into the Moscow front-office and the St. Petersburg office of development and implementation helps us a lot in terms of competition for personnel and allows us to keep the roots.

We recently watched startups who came to Sberbank for investments. Among them, almost half of the participants were representatives from St. Petersburg universities - ITMO, Polytech. This suggests that the St. Petersburg school is extremely strong in the direction of new technologies - high load development, as well as data processing and data mining. As a Petersburger, this causes me great pride.

In St. Petersburg and in Moscow, we regularly organize and participate in SPBDSM - data-science meet up. This is a very cool event for the industry, when representatives of this new profession - Data Scientist - share their experiences. We purposefully invest strength and resources in St. Petersburg in order to create an environment where the guys take the first steps, learn, see what is on the market, communicate with each other. Now the organization of these events is done by people who, including in our team, created the “brains of the system”.

In Hackathon DeepHack on the creation of question-answer systems (Question Answering) on ​​the basis of artificial intelligence, which took place in '16, our guys, who created the core of predictive technologies Segmento, took first place in Russia. And we also competed with representatives of top Moscow universities. Well and most importantly - they took 7th place in the world for the same task, in my opinion, very cool (there were 700 places in total).

We passed three rounds of investment, and by 2014 we came up with a company that was among the leaders in the programmatic segment. By the time we had more than 500 advertising campaigns, we started working with all the largest advertisers.
And even then we understood that the technology we are growing more and more converges head-on with such giants as Google, Yandex, Mail.ru. We had to make a decision: either we are trying to compete only with the efficiency of the technology, or we are making a qualitative leap forward to move to a new level of business.

Although the transaction with Sberbank occurred in 2015, we actually did it for a year - starting in March 2014. The bank was not the only investor with whom we spoke. The key asset for us was in the form of data (they said many times: “data is new oil”) that we could get from a new investor partner.

, , . , , – . , , , . . 70 . , 100, , , . , , , .

( ) , , ( , , ) , . , , – , , , , , - . , , , , , , .



— ? , – , , . , , : , , , . ? , , , - .

-, 2014 , - , , , «Google Translate» : « ». , : « , , » « , ». . , , . . ( ) .




, 2014 , . , , , . , - . , , .

, , , « », , . , .

, , , .

, . , , . , . , . , ( ). , , .

-, , , , , , ( , ) , - digital. , . , , : , Johnson&Johnson «» Johnson&Johnson.

: , , .

, , data mining, — , machine learning, , , , . , . – .

, , . , 100 , .