Geekly Articles each Day
📜 ⬆️ ⬇️

“When you tell a true story, they believe it much more” - Interview with Oleg Shelaev, part 2



A few months ago, we met with Oleg Shelayev , Developer Advocate of ZeroTurnaround (hereinafter - ZT). Since then, Oleg managed to become Java Champion (congratulations!) And ... leave ZT. Yes, if you postpone the interview indefinitely, different things can happen to the hero.

The interview was long, so I broke it into two parts. I published the first part on Habré back in December, and now it’s time for the second part. In it we talked:
')

Here's a video:


For those who prefer to read rather than listen, below is a complete transcript of our conversation with Oleg. Enjoy your viewing or reading.


Try and think


- As one of the great ones said (John Wanamaker - auth. Note) : “Half of my advertising budget is wasted, the problem is that I don’t know which half”.

- It is very difficult. For example, I am writing a post for our blog. Maybe I should write more posts, maybe we don’t need to do time-consuming reports.

- Maybe you need to shoot a video, but not to make posts?

- It is very difficult to understand. Most importantly, when you just try, you think: “Now I will record a video, and, based on the number of views and clicks on the link to the product page, I will know whether to make a video for me or not.” So you will not get this data. You need to do these videos regularly, for some time, only then you can watch and say that you have increased traffic, conversion between the page with the video and the product page, now it works, but at first you can’t say for sure , since there are a lot of factors involved. Most likely, what you started to make educational videos helps you, but you do not know for sure. The only way to find out exactly how much money some kind of initiative brought you, like training videos, is to stop doing this and see how much money you don’t get.

- We do not live in a closed space, we live in a system in which there is a lot of external noise. The effect that you will see from stopping something - the fall / growth of some sales, in most cases leveled by other influences - an earthquake, the World Cup, or something else.

- Of course. But when you start to do something, you have this effect of noise will be much more noticeable. You started to do something, you do not understand whether it is right or not. When you finish doing something, you can clearly realize: we show advertising banners on the site or somewhere else, for which we provide it to companies that pay money. More money - more advertising, more traffic, more shopping. Less money - less than anything else.

- There are complicated things. There is latency - the length of the transaction cycle: how much time has passed since the prospective customer read the post or saw the advertisement before the sale. And if I turned off advertising now, stopped writing posts, it is not clear when I will see this result. For this, I still need all kinds of metrics, such as the average length of the transaction cycle. Is this all considered?

- All metrics related to transactions and sales are considered. It’s impossible to calculate how much time passes from the moment I read the post to the moment when my company bought me a license. But traffic metrics are very simple. I follow them myself, as I supervise our blog, I try to monitor the traffic, I mostly use banal Google Analytics and some Webmaster Tools to understand what traffic goes where, keywords, what people are looking for, what rank we have in Google for some requests. Further it is not converted.

From a marketing point of view, we consider some metrics, some transactions are associated with transactions in the sales department, but basically everything is taken for granted. We have our own metrics. We know that when traffic grows, it’s good, we are great. In principle, this strongly echoes the mission of “awareness” - “brand awareness” and how many people visited us. When few people come to our site - this is not very good. Directly through the efforts of our Developer Advocate, we do not sell. We obey marketers. In many companies, the Developer Advocacy teams work side-by-side with product teams, with people who say: “Now we will do feature X.”

- Are you talking about JetBrains, where is PMM (Product Marketing Manager), or not?

- This is about JetBrains, this is about companies that provide platforms, a cloud or something else.

- That is, it is transmitted by a person deeply incorporated into a product that simply knows what the development team lives.

- And they have metrics and key indicators that can be directly embedded in product development and go along with the development team. For example, in time to tell about all the new features. The development team rolled out a new feature X. Product management said that we need this feature X, the Developer Advocacy team told everyone about this feature X. If, after 5 months of X, 50% of our customers do not use the features, then everyone is to blame. And there the Developer Advocacy team basically works precisely as an aid to the product, in communications and in concepts. And they have common metrics. We have metrics slightly combined with marketing, but not with the sales department.

- There is a well-known joke: “You are a marketing team, you get money for incoming calls. Tell me how much you have from each such call, pay 30% of this amount to us, and we will call you on the desired number in the right amounts. ”

- Almost any metric that is not tied directly to money or to some tangible things is deceptive enough. For example, I write posts for a blog. I would like to write technically literate, deep posts about interesting things for X thousands of people. For example, how to write a Java agent and value types with new byte-code instructions, something deep is straightforward. Cool? Awesome post.

- I would read with pleasure.
- Reads this post, say, a million users, and it will bring me 100 thousand to the site. It's great, I'm good. If I write a post about how to connect the ID generation strategy, sequence, in Hibernate, on the Java side or not, ten million people will come to me. If I have a metric - it's just traffic and some kind of brand awareness, then you need to generate traffic in such a way as to focus more on Junior programmers, to describe some problems of interest to them.

- In terms of sales funnel, there is reach and capture. This is the reach.

- And from the point of view of capture, even if I write very interesting posts about some low-level things that are interesting for more experienced developers, how many such posts do I need? How many such developers still do not know about our products? How many such developers will find out about them and be interested? After all, do I need to write them as often as I would be interested in writing them? Writing some banal jokes is not interesting.

- Yes, and this is also not interesting to anyone.

- No, it is interesting to read. People googling and come, out on Stack Overflow. There are many, you know, very deep and interesting questions.

- Well, they, rather, solve some particular problems.

- They have about fifteen percent of traffic - this is “How to get out of Vim”! They recently published their statistics, and this is a big number. Writing about it is not very interesting, it is difficult to write about, and it’s not quite clear why. On the other hand, you get these traffic metrics.

About RebelLabs


- You have been writing posts for your RebelLabs blog for several years now. Are you tired?

- Before I write a post, I need to study some subject. It is not entirely reasonable to write from the bay-groats, because it will not be interesting to anyone. There must be some meaning. So that people learn something new. It is difficult to write regularly. This is a profession. How many writers do we know? Everyone at heart is a writer, but there are not very many successful writers. It's hard. Playing metrics is much easier, but what's the point of this is not very clear.

In some other companies, where the product is easier to track, you know who logs in, where, what things they use. Clients come to your servers and do something there. It is a little easier to link in a chain. You have a gap between the virtual world, where they sit, and the physical world, where they try your products and do something, then this is again a virtual world, where you communicate with them, and again physical, they sit at the firewall and do not send you have no information - there is a break, and it is much more difficult to watch.

You asked which metrics we monitor - this is traffic, here I’m wondering, basically, what part of our traffic are organic visitors, I’m looking at which posts have been gaining good traffic lately and I understand what post there was that attracts People, I can return, correct its text. Say, in the introduction to write more clearly what it is about, correct SEO tricks. Well, not tricks, there just need to write clearly and clearly, so that even a Google robot can understand what this post is about and what is the use of it.

- Yandex was very interesting about this topic: if ten years ago, SEO and other tricks for search optimization were popular, now they recommend writing content on the site as clearly as possible, and their search engines will figure it out.

- And they are very well versed. But my experience suggests that, if you look at the data, there is also a distortion. If a web framework comes to me on a specific page and does not log in to the Java framework, then I will not have this information in the metrics. I can improve some existing ones, but it is quite difficult to do from scratch. About what Yandex and other similar companies say - “write as it is, don't invent”, most likely they say that if you want to capture a segment, you do not need to start by inventing keywords, creating artificial text to attract visitors. Write a good post in which it will be clear what and how, what benefit the reader receives. And when users come there, when the post becomes popular, when links are sprinkled on it, then the robot will figure out how good the post is and where to put it.

First place in Google converts very well. If you are sitting in the first position for some kind of request, you can see it very well: impression - how many times it is shown, click - how many times you poke at it, and you will have 60% of clicks from impression (click through ratio). If you are in fifth position, then you will have 13%, which is already a very big difference. But on the other hand, you know that you have written good content and you want to be clicked more. And you look, and there sits the Coca-Cola website in the first position. And you just can not beat this site. Or Wikipedia. Against Wikipedia you will not trample. There are limits. You can play with these metrics, but all this will again rest on whether you write good content, you do useful work not just for yourself - as you told the joke, people will call the number and leave their data, and you will generate, you will have a lead generation above the roof. Does it work?

- Do you communicate with other grocery or Internet companies in terms of how the sales department, marketing work, who has the top and bottom sales models? Or are you just trying to read more about it? Let's say your investors say: here it is, and here it is, and communicate with those with these. Or is it more Google?

- I think, at those levels where it makes sense, investors, of course, help. Personally, I find it somewhat difficult to judge, since I am an ordinary Developer Developer.

- For some of our friends, for example, for Hazelcast, JFrog and so on, the level of what is happening, the patterns of how this all happens - are they similar to yours? How typical is your model for this environment?

- I think quite typical. At some point, almost all companies that produce software began to move from licenses and product boxes to a subscription. At some point, JetBrains switched to a subscription-model, we did it a little bit earlier than they did, but they were great, they had a very simple concept of what they wanted to do: to simplify their expenses for operations and logistics inside, to make it easier. They simplified this, and we, too - at some point, our products were sold, say, for half a year. This is not very cool. You have some time to conclude a deal, and then it becomes unclear what clients you have for half a year, which for a year. When you have everything going in one year, it all becomes simpler, you have increased predictability, so we stopped selling for half a year.

About products and services


- And it seems to me, here is an important point - I'm sorry that I interrupted - the company was grocery and was selling some product, say, JRebel, IntelliJ IDEA, and now it looks more like a service model: we sell not a specific product, but we we sell some service that your business will provide for some time with some function. Let's say you can reload classes on the fly in your company without restarting the server. And we are no longer providing JRebel, but some kind of service.

Don't you think that there was any such transition? After all, there are food and service companies. The product is some kind of joke that you can take, touch, poke your finger. And the service (service) client lives .

- On the one hand, yes. But this is not quite right to say. We are positioning ourselves as a grocery company, it is not a service. Roughly speaking, we sell the bytes that make up the license.

- This is in terms of delivery to the customer.

- But this is exactly what comes, they download everything else, everything starts up on its own. In terms of the functionality of the service you mentioned, the service we sell is that our product starts. Without a license, it simply does not start. They say that you are selling not your product, but the vision of how cool your customer will be when he uses this product. Yes, we sell this experience. For example, with JRebel it is reloading classes on the fly, then, as the programmer will be more productive, will be less distracted and, accordingly, save you time and money. Customers buy, use, they are satisfied, because it is so. This is an example of good marketing when you tell a true story, and they believe it much more. But calling it a service would be a bit wrong.

- It looks like some kind of mixed model. In terms of licensing, it is clearly how the service works.

“In terms of licensing, it's just a one-year subscription.” Do we provide some kind of service and do we have any costs? Do we launch our cloud somewhere in the garage and something comes to these computers? Not!

- Well, this is another service, you still only sell resources.

- Our products are in boxes.

- Then such a question. When I was taught marketing, they explained to me that there are several mandatory points that B2B sales should have. For example, people need to understand what they are buying. From this point of view, you are doing a lot of information work. Also, you should be able to answer the question of how your technology differs from competing ones, for example, from Hot Swap. By the way, for JRebel, who are your competitors besides Hot Swap?

- There are some frameworks with built-in ability to reboot itself. It started a long time ago. There was an agent Spring Loaded, then there was Tapestry 5, a Play framework, which is all reload. They all have their flaws.

- Play went on the model of Ruby on Rails, as I understand it.

- Basically there is just a functional replacement of the entire application and deinitialization. Their main disadvantage is that they are sharpened for any one specific platform. "Use our technology, and then everything will work for you." JRebel's biggest indirect competitor is, of course, test-driven development.

- Not a product, but an approach!

- Yes. As most failures are due to the use of TDD, people do the tests ahead.

- That is, roughly speaking, your war for the souls of developers is being waged with Kent Beck and Martin Fowler, and not with Oracle, IBM, and so on.

- This is not really a war. This is a very difficult technological product, it is difficult to make, we wrote it for ten years. If you collect a thousand programmers, they can write it in a year. So, of course, does not work. But it must be supported, have some kind of expertise, it must be lived in order to know how it works.

This is all very difficult, so we have no direct competitors. The teams that tried to do this - more often simply die, because it is difficult. It would seem that it was possible to build such functionality into the JVM, but this is not so simple and not very necessary for someone. And all the frameworks that do this, they have a lot of focus on production, because they have to be good there. Accordingly, the focus on the development is already less.

- That is, the entry threshold is the first thing that helps. The second is the niche that JRebel works not in production, but during the development phase. There is a conscious narrowing of the scope. Like, we are not about everything, we have such a specialization.

- To develop, we have a lot of expertise in the specialization: what we can do, how we can do, how much freedom we have on the developer's machine. To change any classes in production on the fly are risks in security: specially trained people will come and give you a shot.

- You are very interesting to answer the question about your competitors, about TDD. TDD is in your subject field, but it is not a product, it is more a function - “write tests first”. The question that potential customers ask, if the answer to the second question is not very sure, is who your competitors are. I will come to Hazelcast, they will correctly call me GridGain and so on. And when the question arises that the functionality is similar, I ask them about their best price. I will try to understand: here is a Grid, here is a Grid - which is cheaper. In your case, as I understand it, you need to answer what is cheaper: use JRebel or use TDD. But TDD is also expensive, in a sense. Have you tried this type of research?

- How expensive is it?

- Well, relatively speaking, JRebel saves here and here. TDD saves too.

- We have a very good position in this regard - we have a product that works well and is very clearly tied to the programmer’s time. You have an application, you know that your server costs two minutes, you want to start it every 15 minutes, you wrote something and you want to see if it works or not - you lose 8 minutes in an hour. This you multiply by the number of hours, the number of developers and get some figure. It is very easy to compare with the price of JRebel. You look and understand that JRebel costs a penny compared to the salaries that programmers have to pay for drinking coffee every 15 minutes.

- When working without JRebel, the programmer is forced to switch the context, to distract from the task.

- Yes, you launch WebLogic, and you have five minutes until it starts. You go on Facebook, in two hours you come back and try to remember what you did. Oh yes, I fixed the bug. This time is very well measured, as is its cost. Therefore, it is very simple for us: when we give a trial, we directly say - here will be the metrics, that's how you configure it and see how much money you save. Here is our ROI. There are other products in which not such an obvious process from using products to benefits, for example, XRebel, where you looked and realized that you avoided any problems with performance. These problems would have reached production, someone would have said that it’s slow here, I would have to come back - you understand that it would be very long and expensive, but you don’t know how much. In addition, XRebel has a slightly different problem with competitors - there are profilers, APM, some other things.



About XRebel


- JRebel and competition with TDD is understandable. How are things with XRebel? There are probably five Java profilers out there. How do you compete with them?

- Directly. When we started XRebel, we explicitly stated that this is a profiler. It was a bit difficult, because this is a new product, there is a matrix of functionality, there are some expectations from what it can do. The profiler should show you tables of hot methods, which takes time and not only. When you do something else, you have to explain that you are a different profiler.

- Explain that others have the wrong approach?

- No, you have a different approach, which is more useful in some situations.

- That is, XRebel - a tool for finding and eliminating simple performance problems?

- Or for less experienced developers to show them some big problems. We are narrowing down the specification of what we do, and from here we can already talk about what makes our product. In general, there are two big strategies for how to market a product. Or you adjust the marketing and sales department to your product or vice versa - you adjust the product to the needs of the market and the behavior expected in the market. That is, you take a certain market, estimate how big it is, how easy it is to get it, how much money it will bring, and you ask engineers to do this and that.

You have to bring the market and your product together. When you have competitors, it’s more difficult here, because you have to be somewhat like them, otherwise customers won’t understand what it is about. If you declare that you want to be in a niche, then you need to match this niche. On the other hand, you need to be different from them in order to say why customers should choose our product, rather than competing.

This is a much more difficult problem in terms of marketing than selling JRebel, because when you have few competitors and have a good product, it’s much easier for you to show your uniqueness and explain that you need it. The situation with XRebel is different: people use it, people are happy, people renew their licenses for the next year. To capture a larger piece of the market, we need to somehow play with marketing, with the story that we are telling. And this is also not very simple. If you tell a story that is untrue, no matter how cool it is, you sell the product once, get some money from the client. But if your story does not pass the test of practice, then the second time you will not get anything from this client. This is the path to a dead end. Therefore, for a product like XRebel, marketing plays a much more important role.

- Managing expectations is a very important moment. Jacob Fayn, our mutual friend with you, talks about it all the time. If you make the expectations of the product too high, you will buy it more one-time, but there will be less repeat purchases. If you underestimate expectations, say that your product does not do anything particularly useful to the customer, then even the first purchases will be small. How do you play with this? How do you find a middle ground for yourself in this situation?

- Simple enough - based on data from sales. In the same way as there is a game with pricing. We know how much a developer’s hour is worth. We present what performance problems XRebel prevents. We know that productivity is important, and customers recognize that if their application is slow, then customers do not like them, and this translates into money.

We know that it is important that the process of releasing the functionality and correcting any errors takes two weeks, you have one sprint, you can not do it faster. (If you can - well done, your correction will be cheaper already). We can roughly imagine the price, we can focus on the prices of competitors and play in some way. XRebel costs a few hundred dollars a year. If you sell at this price, if you see in the reviews “no, we will not buy your product at this price, because they are too high”, then this is a price problem, and you have to reduce it. If you do not encounter such problems, it shows that the market is ready to pay, just your product is not yet at the point where it wins over the competition in functionality.

- On any five potential buyers there is one who will say that he is too expensive. Always, at any price, there will be people who will say that it is expensive, and there will be people who do not care how much to pay. It is clear that finding the optimal price in this balance is a rather difficult task.

- It's complicated. You can't change the price every week. It is simply incorrect, even impolite. This is a difficult process, both external - to explain that we are now changing the price, and internal, where you have to restructure some processes in the team, because you have changed the price. But it needs to be done.

- I like your marketing idea: telling potential customers how much time you save them, because this is an attempt to rationally explain why it costs so much. An attempt to come to the market and say: here is 500, and here is 600, so our price is lower than that of competitors - it is an attempt to show money that a company or team will save on using your product, not your competitor's product.

- This is an attempt to explain the value to the client and to describe the vision of what he actually buys. If the client did not like it, he, of course, can zapapadadit this jar-file us back to the server!

GeekOUT and VirtualJUG


— . -, GeekOUT — , , . : . — Java -. , - Duke's Choice Award. , JUG', , . Java User Group, .

- Virtual JUG is a Java virtual user group, it is official and not owned by ZeroTurnaround Corporation. This is a voluntary organization. The idea appeared in 2013. At the source was Simon Maple. Looking back now, it seems obvious that it would be very cool if the concept of a user group with regular meetings and interesting content from speakers would be produced online. Then we thought, why don't we do it and not try.

The technology is quite simple, because now video conferencing is easy to do. We use the banal Hangouts, this is all automatically broadcast on YouTube, and people just come and watch the video. We have a chat where people can chat. The biggest bonus is that we have small expenses for organizing even world-class speakers, because they do not need to fly anywhere, they connect from home, tell their report, communicate with the audience, meet with us, and meet with minimal costs. People liked this concept very much, we began to grow fast enough. On Virtual JUG, we have about 11,000 sympathizers — meetup.com subscribers. We registered VirtualJUG as an official mitap, people “come” there, we post a schedule of future reports there, they can respond. This is the basewhere there are emails and a list of registered people. I don’t know how many YouTube subscribers we have, it’s a secondary metric.

- How does ZeroTurnaround integrate with VirtualJUG? I saw some variants of spinning ads on YouTube, using the email base of subscribers, as well as native advertising, when inside the video Simon Maple talks about some new feature of your product.

- ZeroTurnaround fully sponsors VirtualJUG, we organize it for the time of ZT.

- So, part of your working time is Virtual JUG?

- Yes, this is support and organization of Virtual JUG and participation in this community. We do not insert advertising into the video, at the beginning of the report we say that the sponsor is ZeroTurnaround, we thank them, they have basic products for Java developers, if you look at them, it will be good for everyone, if you like it, great. We do not use the list of participants for the promotion of our content.

— . sales email-?

- Of course.There is a joke about how an aggressive Estonian seller works : he comes into the office in the morning, sits down and begins to aggressively wait for a call in order to make a deal. ZT started out as a small company, we are very closely connected with the community. In order to be successful in the industry as a whole, we must have a large community, where people are happy to learn, work, learn something new and share knowledge. We are happy to support this. VirtualJUG is a separate organization sponsored by ZT, we do not use the list of its members when solving ZeroTurnaround sales tasks.

- Now we need to conclude an agreement with the client about who and for what purposes will use his personal data. In Russia, this is a very hot topic.

- I don’t even know if we can technically download the list of participants c meetup.com download in Excel and put in Marketo. We can write emails to them, periodically doing this when ZT conducts webinars, for example.

- You can more accurately communicate with the audience: let's say, look, look, an interesting article was published on RebelLabs of this speaker, who we had two months ago. That kind of thing.

- Or, for example, when we did some review of the industry and sent a survey to the participants of Virtual JUG, asked to answer, because we know that this is the target audience, and they will be interested. When we make it, we will also send them a report. But there is no direct monetization.

About simple and complex chains in marketing


— . , , Virtual JUG . , - . , , : , Virtual JUG, RebelLabs, , , , RebelLabs, RebelLabs, .

. : , ? , , . ?

- In my experience, not very helpful. Yes, theoretically, it sounds very good. We have a six-step Journey:

Unfortunately, the situation is the same, even though Virtual JUG has 11,000 people, which seems like a fairly large number. Actually, if you look within the framework of a funnel, this is not very much. If you are eliminating 50% at every step, then you just have a stunning conversion. With such a conversion, very few people will come to an end. And you can do it once.

- That is, you will come to 100 with something. 64 times you lose.

- This is very small. This is a conversion for which any marketer will give an arm and a leg.

- This is not the coverage that you can work with?

- This can be done if you are a Coca Cola volume company.

— , . — , . , , , . — , , .

- Yes, and it works. If you build content well, and at every step you will have interesting information that gives your reader and visitor something, it will work. Is it possible to build a business on this in the framework of the Java-community? I doubt. In my experience, if your post consists of two parts, they click on them. Of the three parts or two parts with follow-up, there is already much less traffic going there. I can use it in the long term, it is very difficult, but we are striving for it. But there are no such results so that we can say that it works and recommend it to everyone. One-way combinations work. I wouldn’t bother sticking advertisements anywhere and everywhere. Frankly speaking, engineers, programmers, team leaders - they are all intelligent people.

- It seems to me that people simply do not like clumsy work, people love it when done finely.

- Yes, all people are not stupid. Remember, we were at JCrete, and there was a group report about self-promotion. And Jacob Fayn told us, they say, let's write posts. And this is very correct, it is a good idea. I’m a little ... well, I didn’t troll him, but rather told him simply about metrics how to play them, optimize them. And he asked me if I thought that all people were idiots and did not see through the transparent system of funnels, that we were showing them A to buy B.

Of course they see. No one can be considered an idiot. I am for simple combinations where you clearly and honestly say that you wrote, but your time was sponsored by such a company, and they are great. If you're interested, take a look. It works much easier and easier than building some complicated funnels.

Here you organize conferences, and you often write articles on Habr. You can look at your metrics: choose a separate popular post that generates traffic well, see how it is converted, whether people go from there to an article about JPoint, Joker or some other conference. But if you try to build a funnel or a graph from existing posts, then it seems to me that you will also come to the conclusion that this may not be worth the effort that you put into it. We need to honestly say: yes, this is our business, and we are trying to do a good job. If you, as Java programmers and engineers, would be interested to go to the conference, here is a great opportunity. And building some ideas is a very complicated marketing.

Source: https://habr.com/ru/post/347578/

More articles:


All Articles