20 product managers and the most multi-dimensional matrix structure of all. Conversation with Skyeng

Recently, the Habra product team began to actively attend conferences. And we noticed one interesting thing - at almost every conference there are speakers from Skyeng, and almost always their reports turn out to be the most interesting. It turned out there are two reasons for this.



First, there are special coaches inside the company that can prepare any employee for a public speech. The second - product managers have gained really powerful experience, and they have tried many experimental methods of product management.



To gain experience with colleagues, the whole team went to Skyeng to visit, and listened to why companies have more than 20 products, how each of them manages the budget and team, and why everything that is defined now can change tomorrow.

Skyeng at My Circle got an average rating of 4.2 out of 5. Judging by the estimates, employees believe that the company is making the world a better place, and they also appreciate it for its modern technology and excellent team relationships.

Denis Pushkin, Head of Product Marketing at Skyeng



- Maxim Karakulov (Product Director in Habré): How is product management structured? What parts is it divided into? Who controls what? Especially intriguing are indicators such as: “We have 40 products and 40 analysts.”



- Denis Pushkin (Head of Product Marketing in Skyeng): There are 20 people, and 40 analysts, yes. But you need to understand that not only products have analytics. Sales, OSes also have them. Some have several analysts. Now there are five analysts in the sales department, because there are several lines. Products in a team have one analyst, a maximum of two, because the teams are small.



- MK: How did you get to this? How did you grow up?



- DP: This is a very difficult question. I really like the book From Good to Great. Its author is trying to figure out what made the great companies such and concludes that the silver bullet does not exist. There is only one factor. We need systematic work, certain principles and attitudes that allow the company to grow.



Our growth was iterative. We just watched what we were doing, what wasn’t working, we learned the expertise from the market, corrected it and went on. I like to think that we grew so fast because many things were not learned from our mistakes, but built a culture of communication with other people and companies. It seems to me that Skyeng has collected the expertise of the entire market. In the area where we work, we know almost everything that we need - but we know this not because we went through everything ourselves, but because we talked a lot. We have heard about all the top practices at least once, and we can apply them.



- Nastya Khovrina (Product Manager in Habré): It is interesting to learn about tipping points and crises.



- DP: As far as I know, the guys had crises and difficult moments at the very start, when they were still living in a hostel. Once it was necessary to pay a salary to teachers, and the money was just a cache. They inserted a card to put the money in the ATM, the ATM took the money, but the money was stuck.



On the day of salary payment to teachers, 700 thousand rubles got stuck in an ATM. That was a crisis moment.



In fact, tipping points rarely occur in such companies. Just a gradual development.







- MK: And the profit?



- DP: There were small investments at the start. But we are paying back. We are lucky there is no competition in the market. In the taxi market, you can’t be self-supporting now. Or in the food delivery market. Yandex and Mail, - two giants of traffic - just wet each other.



And we do not need to compete for marketing with Mail.ru, which has free traffic, and they just flood everything. More traffic than we in education is unlikely to be purchased by anyone.



- MK: Have you always had enough funds to expand?



- DP: Yes. We even had investments from Baring, but they are on the account, because they are not needed. We could develop even faster, but so far we have no time to cope with the current growth. Today I will talk about one thing, and tomorrow everything may already be different. Everything changes every day.



We have a matrix structure, and to our horror, it is already multidimensional-matrix. The product team has a product manager. He is the king and the god of the team - owns a budget, manages people's salaries, plans and so on. The product manager (depending on the team) has developers, QA, designer, analyst. There are about 20 such teams, about 10 of them each. More than ten, probably only in the CRM team, but it is also divided into subgroups.



This happens when teams grow and mini-teams separate from them. For example, I was an ordinary product, then I took more and more directions, I could not cope with them myself, and I hired more products. I gave them a piece of the team, then increased it again, and again gave it to a new product.



Teams have the ability to clone, disband, grow. They have every opportunity. But if the project brings little money, is unprofitable or uninteresting, we can disband the team and distribute people to other teams. Do not disappear to the same people.



There were cases when the team closed one project and in the same composition went to do another. There is a product manager in each team, but above them there is a senior product who leads several product teams. He is responsible for the final result, and pretty much immersed in the metrics - not what the guys are doing.







- MK: But does he see concrete cases?



- DP: Yes, he can get into concrete business, but basically he tries to abstract himself. Specific products are run by juniors and middles. Signors lead several projects at once. There are not many of them in the company. For example, the product lead of sales teams, and he now has three product teams under his command.



And there is a head of product that can control horizontally and vertically. They have subordinates, and leads and ordinary products that do not yet have their own leads.



What is the difference between horizontal and vertical submission? Vertical submission is when a team is within your budget, and you practically command it. And horizontal - for example, I have a team, group lessons for children. Vertically, she submits to the leader of the vertical of children, but horizontally I “mentor” her, because I have expertise on their issues.



I ask the result from some guys, but I don’t ask from others. My main job is to build team processes. This is a horizontal obligation, and if work is poorly structured somewhere, the wrong people are hired - this is my fakap. But I am not responsible for the final result of the work.



The rest starts generally interesting. The product manager has guys who are involved in different types of work. For example, developers. In order for them to have cool processes, their horizontal leader appears. In an ideal world, he is directly subordinate to head of development. But we now have a complicated, complicated system, so it is subordinate to head of product.



There is a technical director, but now the most peppy commercial products are not subordinate to anyone horizontally. They are on their own, and techdir does not go into their lives.







- MK: At what level do your heads manage? Are they abstract from specific projects?



- DP: They can be playing trainers in a bottleneck. If they see a problem in some team and understand that the team, for example, is a fig team leader, then they decide to hire another. Sometimes they are very deeply immersed in processes. They say: "I have been in the team for a month, do not wait for me." They begin to build processes for them, hire leads, train them and then go back.



Everyone has their own decisions, their own management methods. Some will first hire a new lead, teach everything, and only then fire the old one. When something needs to be done urgently, heads get into projects with their hands. I myself have several projects that I practically run with my hands. But this is bad practice, you have to leave it.



It takes a lot of time, and the effect is sometimes not very large. Spending an hour on Lida is more effective than on June.



- MK: I saw a cool presentation of such a scheme. There, all the teams were painted like slices of pizza. For example, each sausage is QA, cheese is design. Accordingly, all the pieces are separate independent products, but they are united by a common horizontal. And in such a scheme there were no heads.



- DP: We once didn’t have them either. There were separate teams, but everything grew. Too many things appeared, they were delegated and new products were hired. For example, I already have 14 people - what should I do with them? I say: "It seems that the time has come to take a few guys."



- MK: Such substitutes are obtained, right?



- DP: Yes, all the alternatives to each other. If I get hit by a car, there is someone to catch the fallen knowledge. And there should be several people. But there will always be a leader to whom to replace, but there are products in the teams.



Making cool teams doesn't always work. Even the most independent teams need motivation and control. A person who reminds them focuses, sets goals, and so on. We did such experiments - not on purpose, but it happened: the product weakened attention, was engaged in other affairs, the results of an independent team became worse. Then the product came back, like it didn’t do anything special, it just asks the guys: “How are you?” And they are: "We have not done anything." And then it went, went, they do it again on their own.



- Denis Sobolev (Product manager b2b in Skyeng): I can add about the horizontal analytics. When a team of analysts just formed, there were about ten people in all of Skyeng. We worked independently in the corporate department. The analytics team has overgrown on its own. I was the first analyst, two months later another analyst appeared, then two more, and the result was a separate team, and the product was our customer. He came to us, said: "we need reports, studies." We gave. But my manager was a commercial director - I worked with him directly.







- MK: But gradually, have analysts become part of individual projects?



- DS: Analysts themselves can be the initiators of some projects.



- DP: I came two years ago, six months earlier than Denis. These six months that have passed between us are just infinity. Then the movement began to saw all Skyeng into units. Initially, Skyeng was one team, one product. When I arrived, I began to unhook the marketing part from a single product. We wrote separate services, allocated separate teams for them. Horizontally began to appear.



Initially, there was one development service. Marketing did not like that all its tasks went with low priority. He wanted his development team. Product marketing appeared, they took me there as a product manager. I began to focus on tasks for marketing. Then everyone also wanted everyone else. And gradually one team began to saw itself into pieces, units, services. Now the idea of ​​micro-units has become fundamental to the culture of the company.



- MK: And what is the role of CEO and all sorts of strategies?



- DP: The company has common long-term goals. For example, by 2024, get a valuation of one billion dollars to enter an IPO. Very simple, without troubles. It remains only to do so.



Further, the guys from above cascade this goal and say: “It seems to us that the adult business vertical can bring so much money, children so much, corporate so much. And we would like you to bring so much. " Planning of the year begins. Everyone will do according to plan what they can. everyone knows the common goal and decides how much he is willing to commit. Then it sums up. From what we have proposed, they will tell us: "You will do twice as much." And we will do twice as much. It is always negotiation.



I have exactly the same process. I say: "I will bring the company 800 million revenue in a year." And then I say: “Guys, how much can you bring?” They are: “100, 200, 400.” I see that another 100 million is not enough, and I say: “Maybe someone else will bring 100?”. They say: “No, it will not fit into us. We are afraid, we don’t want, and so on. ” Okay, it seems it's time to launch a new direction. I get ideas for new directions and look where there is the potential to bring another hundred million.



Usually I do this not in fact, but a little in advance. In order to bring 100 million next year, I am already launching a direction that will reach full speed next year and will make money. Because it's still time. Products do not immediately start to make money.



- MK: When does the specifics begin: “What are we going to do, what to develop”?



- DP: There is also a horizontal-vertical position here. When you are a horizontal leader, you just walk around and sow your ideas - suddenly someone will sprout. If I’m a vertical leader, I can say: "Guys, this is a strategically important area, direct your focus here." But the guys still achieve the main result.



They do research, look for growth points. Based on the found they say: "Here is the plan, we will bring so many millions." And I can ask them: “Are you sure that this is just so much? Maybe twice as much? ” And they can say: "Yes, we are so sure and twice as much will fail." Or they can: "No, okay, another 1.5 times is possible."







- MK: What about products that are without teams?



- Matvey Matveev (Product manager in Skyeng): For example, I have a new direction - I am a product of a referral channel. This is an important product, but not a strategic one, it does not have a Green light tag.



- DP: Green Light is an opportunity to do cross-team projects. Everyone has their own metrics, and China will come to me with its task. I’ll say I have KPI metrics and I’m not up to you. And if there is a Green Light label, the product gets end-to-end priority through all the teams. At our annual annual session, Green Light projects are determined.



- MM: In order to implement referral I need marketing resources, developers. I go, for example, to the product manager who is involved in the development, I say: “Listen, I have such tasks. We need to cut the banner, because we want to test. Here’s an epic, here’s a calculation, let's think about how to shove it. ” And we are thinking how to shove it.



Or, for example, the product is responsible for marketing, but in part it also deals with referral, so you can call it submission, but still, we do something together in a team. For example, I am responsible for my personal account. My tasks are not strategic yet: something needs to be done in the mobile version. And then China comes to me, which says: “We urgently need a mobile version, because there is mobile first, we need to translate into Chinese, we need a referral, because in China everyone loves to invite friends.” And with these Green Light, I come again to the product that is developing, I say: “Listen, here is Green Light. Let's push these projects, tasks. ”



And here a very funny thing begins, because in addition to the Chinese Green Light, we also have Green Light. And we begin to measure whose Green Light is a priority. Naturally, we’re trying to agree, or, for example, speed up the search for developers,



Each team also defends projects for the next quarter. I come to the defense, and I understand that we wanted to do about the same. Then we discuss how we better share resources and interact.



- MK: Why not assemble your team?



- MM: While we are collecting it, we must do something. I have to pay for the team. If I then hire a team, I have to drag such projects that will not just pay for it, but give multiple growth.



And, if we understand that we will bring a lot of money and we need to invest very well there, then I say: “That's it, we are building a team.” In the meantime, I worked a little here with the analysts, somewhere with the research team I worked. In China, for example, the guys and I agreed that they would give me their scammers, and I would horizontally manage them, what to research, where to look, and they would already explore the Chinese market for an example of how referral programs work there.



- DP: We have such a practice that if you need something big from a neighboring team, then the neighboring team can fix a resource for a fixed amount and give it to you for management.



For example, we have a project of introductory lessons. A very important metric for them is the number of technical failures when entering the introductory lesson. And one team gives another developer. This allows you not to hire in excess of resources, not to cause chaos.



At the end of the month we issue team accounts to each other. We have a special file that is written to. If we ourselves are developing something for CRM, then at the end of the month we will set the development cost for them. They will take it to their budget and will have to pay back.







It started with marketing. They needed integration with a new partner, but in the end they did not start, and we spent 300 thousand on development. And the next time they come, I say: “OK, are you ready to pay 300 thousand for integration?” They will refuse, because their KPI depends on the cost of attraction. If the cost of attraction grows by 300 thousand, and they take zero out of this, then, of course, they will be left without bonuses.



This allows you to balance, more accurately order the help of neighboring teams. They think it is necessary or not, they will use, they will not use. Sometimes it’s more profitable to take external decisions. Why should I spend on resources on the little things if someone has already done this.



If we hire a developer, then HR invoices for hire. In essence, such recruiting is an internal agency. If the recruiter works poorly and for a long time, I can even go to an external agency. If I ordered a developer on the market, then this will go to my budget. If a team needs photoshop, they pay a license for Photoshop from the team’s budget and simply pay for it. If I need a sweatshirt for a speech, then I go print it myself, issue an invoice, and this money falls into my budget, and I have to pay back on sweatshirts with my work.



- MK: This, apparently, is related to how you set goals in money?



- DP: Including, yes. Different teams work differently with us. There are two approaches to global product teams - OKR and KPI. Commercial products work according to KPI, because they are very easy to count - we have quick conversions, first payments and clean money. Our KPI is the net profit we earn minus the costs of the team.



For example, I say: "I will bring the company 100 million profit, and spend 10 million of the budget on it." That is, my KPI is 90 million.



- MK: And do you even include sweatshirts in expenses?



- DP: Yes, of course. It will be 10 million 2400 rubles. Matvey says he will bring referral 30 million, but for this I need to spend 5 million. , ROI , ROI 900%, 90 10, 9 900%. KPI 25 ROI 500%. , - ROI.



, . , , . . , . . : «ROI 200%». : « 200, 700, 200, 500». . , , ?



— : .