How Anti-Retention Metrics and Next Best Actions Boost Retention

Martin Gonto


Co-Founder & GTM Advisor


HyperGrowth Partners
Martin Gonto
Martin Gonto

Episode Summary

Today on the show we have Martin Gonto, the Co-Founder and GTM advisor of HyperGrowth Partners and interim CMO at Vercel.

In this episode, Gonto shares his experience in identifying and leveraging anti-retention metrics to optimize product strategies.

We then discussed the concept of next best actions and how these strategic steps can significantly enhance user retention.

We wrapped up by exploring practical examples from Gonto's work with various companies, highlighting actionable insights for startup growth.

Mentioned Resources



Introduction and Welcome00:00:34
Working with Vercel: Engagements and Learnings00:04:13
Impact of Generative AI on Product Development00:05:23
Anti-Retention Metrics: Concept and Importance00:07:00
Real-world Examples and Case Studies00:09:38
Next Best Actions: Implementation and Benefits00:12:35
Revisiting and Updating Anti-Retention Metrics Strategies00:20:53
Final Thoughts00:33:13


[00:00:00] Martin Gontovnikas: Find a way where you can track when and how people are using every feature. Then the idea is that you start plotting how people use this feature throughout a timeline. So it's like, okay, people sign up. At what time do they use feature one, two, three, four, five. In our case, we have a timestamp in the data warehouse every time that somebody used a given feature, at least for the first time.

[00:00:34] Andrew Michael: This is, the podcast for subscription economy pros. Each week we hear how the world's fastest growing companies are tackling churn and using retention to fuel their growth.

[00:00:34] VO: How do you build a habit forming product? We crossed over that magic threshold to negative churn. You need to invest in customer success. It always comes down to retention and engagement. Completely boosts the strategy, profitable and growing.

[00:01:00] Andrew Michael: Strategies, tactics and ideas brought together to help your business thrive in the subscription economy. I'm your host, Andrew Michael, and here's today's episode.

[00:00:34] Andrew Michael: Hey, Gonto, welcome to the show.

[00:01:13] Martin Gontovnikas: Thank you for inviting.

[00:01:14] Andrew Michael: It's great to have you for the listeners. Gonto is the co-founder and GTM advisor of HyperGrowth Partners and interim CMO at Vercel. Gonto is a growth advisor to startups such as Retool, Deepgram and Ashby. And prior to HyperGrowth Partners, he was the SVP of marketing and growth at Auth0. So my first question for you today is what was the inspiration behind HyperGrowth Partners that you and [G] co-founded?

[00:01:37] Martin Gontovnikas: The main idea around HyperGrowth Partners was that we both had a lot of experience around helping in GDM different companies. And we wanted to be able to help and work in multiple spaces at the same time, because it was fun for creativity, but at the same time, we wanted to get some upside out of it. So we wanted to charge equity. So if we helped and they grew, we basically would get more than just the cash from helping the companies. That's basically where it started.

[00:02:08] Martin Gontovnikas: And then eventually the idea was to collect basically a lot of partners. We now have 40 partners that are working together, basically trying to help them of these companies and trying to find a way where we can match make C levels with companies that are maybe not the station they could hire them, but where they do need some of the growth and GTM help.

[00:02:29] Andrew Michael: Wow, that's amazing. It's interesting how fast it's grown as well. Obviously we had G on the show previously, also it's still one of the most popular episodes, one of my favorites as well. I think people, as you mentioned, calling them mad scientists for a reason. And it was very interesting. So, but it's really impressive how fast it's grown over the last couple of years. Like what's the plans now going forward?

[00:02:48] Martin Gontovnikas: So for now, we want to continue to focus on both marketing and products. One thing that we're thinking about for the future is do we want to get eventually into sales? I don't know. But the main thing is we never want to get more than between 20 and 30 companies a year, which is where we're at now. What we want to do is improve the quality and the caliber of companies that we get and of partners that we get.

[00:03:11] Martin Gontovnikas: But where we do want to focus on in the future is creativity, which is still where we're focused on now. Like every time we meet a new company, they talk to us about, okay, what are the frameworks that you implement? And my answer is always the same, which is like, if frameworks will be successful, everybody will implement them and they will be extremely successful. So it's not about the framework. It's actually about deeply understanding the problems and finding unique and creative ways to build a GTM strategy around those.

[00:03:43] Andrew Michael: Yeah, it's interesting. I think it's like that other quote you're saying. It's like, if there was such a thing as a shortcut, it would be called away because who wants to go the long way around? But as you sort of say, like you need to really figure out for your business. And I think that was like the big premise of the show is all in the beginning was that.

[00:03:57] Andrew Michael: We talked about it before the show, like the seven friends in five days from Facebook, there was like the silver bullets and everybody thought there was the solution for their business, but really it's about figuring out your business, your space, your customers, your go-to-market motion, like how you can solve it. So nice.

[00:04:13] Andrew Michael: And then typically like, how are you working? You've just joined Versal now as interim CMO as well. Like what would a relationship like that look like for you and how long do you expect to see an engagement like that?

[00:04:24] Martin Gontovnikas: So typically we work with companies two to eight hours a week, and we do contracts between one to two years. There are certain cases where they run so fast that we've renewed because the help that they need, the teams that we work with are different. Vercel, for example, is one of those cases. Ramp is another one of those cases. And being this interim CMO is not something that we typically do, to be honest, but it's something that we're exploring now to see how it goes and what that means.

[00:04:55] Martin Gontovnikas: But typically, as I said, our insurance are two to eight hours a week working with somebody in the company. So we need somebody in either marketing or product that we can partner together and we can tag team and work on that. So they can basically execute on our advice.

[00:05:11] Andrew Michael: Very cool. Yeah, Vercel as well is an incredible company, I think, at the same time, seeing how fast they're growing and what they're turning out as well at the moment. It seems like they're a very interesting intersection as well with everything that's happening in generative AI.

[00:05:23] Martin Gontovnikas: I'm a big fan of BZero, the product that they just launched that allows you to either upload a picture and get the React code or be able to chat with it. Like that to me is mind blowing. So I think they're in a very good place to lead Gen AI for developers.

[00:05:39] Andrew Michael: Yeah. And I think it's become incredibly powerful as well. Over the last month, I've started building my own product again. And I would not call myself technical. Like I knew how to do a little bit of front end work previously. But just using things like ChatGPT, I've been using another product called Cursor. Like I've built an end-to-end platform within the last month, which probably would have previously taken three engineers and six months worth of time. And it cost me $120 so far in GPT-4 credits. I got a notification that I'd reached the limit two days ago, but it's just incredible how fast and how far it's going in such a short space of time as well.  And I think Vercel, like as I say.

[00:06:19] Martin Gontovnikas: I was an engineer in the past and I wanted to test these new AI tools. So I was like, okay, in the past, I've coded React, but I've never done Vue. So I ended up using Sirius, like Cursor, GitHub Copilot, and Visio to start thinking about how do I prototype an app? How do I think about this? And I was able, with my programming knowledge that I haven't used maybe in 10 years now, to still code an app and do it. So it helps me accelerate. Imagine like junior people are some of the other people that are coders in the world.

[00:06:50] Andrew Michael: Yeah. I think the acceleration all around from software is just going to be exponential now over the next few years. It's going to be amazing what people are coming up with left and right. Definitely an exciting time.

[00:07:00] Andrew Michael: So let's dive into the topic for today. I was here, like the show, Churn FM. We're going to focus on churn and retention. And at the start, you mentioned something around sort of like the anti-patterns or anti-retention metrics. So maybe you want to just give us an overview of what this is and why you think it's important.

[00:07:18] Martin Gontovnikas: Yeah, when I talk to founders and I talk to companies, everybody comes with this idea that now I think people in growth get it, which is like, okay, I need to define an activation metric, I need to define my retention metrics. And then if I drive towards those, everything will work. And I think now all growth people think about that. But I think there's very few people who think about what are the things that can actually break retention instantly. There are things that a user can do. that might break retention if they do it too early or if they do it too late.

[00:07:51] Martin Gontovnikas: And those, I call them anti-retention patterns. And we actually discovered them first at Auth0. When we were doing analysis of people that were retained, what we did was we looked at people who, after signing up and being activated, in 12 months later, they were still active with 100 users, comparing those that weren't active with those 100 users. And we started to dive deeper into some of them.

[00:08:19] Martin Gontovnikas: And what we saw is that a lot of the people who became activated, but then eventually lost retention, had actually tried to implement a feature that was very hard for their stage too soon. I'll give you an example. One of the things that happened there was that they tried to implement multi-factor authentication. And implementing multi-factor authentication is something that is not very easy and it's not a concept that a lot of people understood, at least back then. Now I think it's something that is a lot more common.

[00:08:50] Martin Gontovnikas: So when we saw that people tried to implement multi-factor authentication in the first week, 90% of them stopped being retained at Auth0. However, if they actually continued to use Auth0 and only tried to implement multi-factor authentication after a month, then that would actually help. So what we started to see was, not only that there were certain steps that people should do to get to retention, but those steps were a bit fuzzy. So it wasn't that easy to understand, okay, is this always the first or this is always the second?

[00:09:25] Martin Gontovnikas: But what was very easy to understand is if there were certain features that were harder to understand, that they implemented in a time frame that was shorter or was before, eventually they immediately dropped and it killed the retention of them.

[00:09:38] Andrew Michael: That's interesting. And I think it makes a lot of sense in the context of a product like Auth0 as well, where there's varying levels of complexity for implementation and things do become more important as you go. So like that specific feature as well, multi-factor authentication, like it's typically something maybe startups would want to put in their enterprise plans or on a higher premium plan.

[00:09:58] Andrew Michael: And when you launch, maybe it's not something that's necessarily needed. So you could sort of see how that plays out. Seeing it, Auth0 then how have you then seen this play out maybe at some of the other companies that you worked with?

[00:10:10] Martin Gontovnikas: So for example, with Vercel, we've also seen this where Vercel has a lot of hidden features. Like you deploy an app and that's it. But then after deploying an app, there's also other features like dev mode that you can add, like certain type of caching and stuff like that. One thing, for example, that we saw with Vercel is that if they used a specific caching strategy too early, again people would drop, for example, and that wouldn't work.

[00:10:27] Martin Gontovnikas: In Humanitech, which is a platform engineering company, they also had something similar happen where there were three product concepts, and if they started to implement certain type of inputs, it also would drop them. So it's been very interesting that it's not something that only happened at Auth0, but in all of the products, there's a certain level of complexity that people need to start understanding.

[00:11:02] Martin Gontovnikas: And one thing that to me was interesting about this was also that as things become harder to implement, not only the past knowledge helps, but also how locked in or how much you're using the tool matters. Because if I just tried it for a day, it's a complicated feature, I'm done. But if I've used it for a month and it's like, okay, I like everything that I'm used, and then this feature is going to take me a bit longer to implement, maybe it's still worthy to do it.

[00:11:30] Andrew Michael: So it's sort of like, I think this is something is all we chatted recently is all with Tatiana about like the concept of commitment curves. I don't know if you've come across them as well, which is another way of like mapping out usage and just as the level of complexity or as the intensity of usage grows, so does the difficulty and level to which they're committed to your product. And I think that's also like an interesting view of it. But yeah, let's dive into the sensor.

[00:11:55] Martin Gontovnikas: I've actually seen of commitments in the past. I just think that commitment curves are harder to understand and measure than an anti-retention thing. Because this is like, okay, you just do like a cohort, you see the features and you know when not to show them. And maybe it's about you hide them or you show a warning before or something like that. But I think it's an easier to grasp and see concept than just understanding the site. Like that is also how like the site framework was very similar to this idea of commitments. But it was very qualitative. And I think that to act on it, you need something that is a bit more qualitative.

[00:12:35] Andrew Michael: Quantitative, yeah, that's Darius Contractor, the Psych framework. I think we also had him on the show previously. But yeah, so let's dive into it a little bit then as well. You mentioned it's very easy to look, very easy to understand. Let's say I want to get started with this now on my startup and figure this out for myself. Like what are the steps that you would take with the team to get to the point where you can say, okay, these are the metrics that we need to be like stalling and letting people find out about them later.

[00:12:59] Martin Gontovnikas: So to get it started, what I would say is first of all is find a way where you can track when and how people are using every feature. Then the idea is that you start plotting how people use this feature throughout a timeline. So it's like, okay, people sign up. At what time do they use feature one, two, three, four, five? In our case, we have like a timestamp in the data warehouse every time that somebody used a given feature, at least for the first time.

[00:13:28] Martin Gontovnikas: And then if you have users, you need to compare users that were retained after a year, closed and weren't retained after a year, and start looking in a Google Sheets typically into, okay, for each feature, which ones were implemented in the first hour, first day, first three days, first week, first two weeks, first month, first two months, first three months, first six months. And then starting to compare all of them, both in absolute numbers as well as percentages, to see a path.

[00:14:01] Martin Gontovnikas: And then for those that were very high in people that were not retained, starting to dive deeper into what happened. And as I was explaining, for example, in the Auth0 case back then, the idea was in this multi-factor, which I think is the easiest to understand. When we're looking at people that had zero, we saw there was a high percentage of people that tried to implement it in the first week. So with that, it was like, oh, wow. So the people who are retained implemented in the first month.

[00:14:31] Martin Gontovnikas: But the people who are not retained is the first week. And once we start seeing the gaps, then you can understand it a bit more. And then we validated that by actually looking at recordings. We've used Hotjar or Full Story or something like that. When we started to look into recordings of how people were using some of the features, and a lot of times we started to see that they were lost or something was happening.

[00:14:55] Martin Gontovnikas: And that actually gave us another idea that we ended up using, which was we started to get SDRs to reach out when people got blocked trying to do something. So by looking at the recordings, we learned that if people click in a section, click in another and then come back, likely they are blocked. Or if they click in a section, go to docs, go back to the section and do nothing, they didn't get it and they don't know what to do.

[00:15:19] Martin Gontovnikas: So we started to also understand and see patterns of people being blocked when they were starting to use some of these features and then eventually leave. So for those who were like, okay, we need to do something to hide these features or disallow them to use them. But if they end up using another feature and they're confused, maybe we can get somebody to reach out and build a relationship.

[00:15:40] Andrew Michael: That's a nice use of recordings. So to recap that as well, then essentially what you're saying is that you're looking at the first time somebody used a specific feature within the product or service, you're taking the average then in terms of like the number on which day it was or which week it was. So you're like, as you mentioned, you have two cohorts, one that retained more than 12 months. One that didn't. And let's say the one that didn't, there's maybe a thousand people you look at, at which day those thousand people used feature X and then you get the, the average and say, okay, like this is the day in which the majority of these audiences doing it.

[00:16:14] Andrew Michael: And then you're looking at the difference between the two and seeing where there's standouts between those that retained and those that didn't. Nice. And how many features then did you see at that point? So obviously the one was like, and from your experience, like how many do you typically see in terms of features? Is it like one, two? Like is there multiple different features?

[00:16:33] Martin Gontovnikas: To give you examples, we typically copy at 20 features because I think looking at more than 20 features is too much. So in a lot of cases we group features that we feel are similar. And we always get it to 20. We've done that with all of the companies that we worked with. Out of the 20, typically three are the ones that are getting people stuck, which doesn't sound much, but it's 15% of the features. It's a lot, I would say.

[00:17:00] Andrew Michael: For sure. And then, so you have this information now, like you figured out, okay, these are the problems that we have at this point in time. What does it look like internally then as well around like going about experimenting with this and like, what are some of the experiments that you want to start thinking about? So obviously you mentioned a couple previously, but maybe we can talk a little bit more in concrete terms as well.

[00:17:18] Martin Gontovnikas: Yeah. So I mentioned a few. One was definitely having this type of like SDRs that were actually, because the products were technical, they were somebody technical that were contacting people when they were stuck to try to get, to help them get unstuck. That actually did work for mostly the enterprise ones, not as much for self service, because for the enterprise ones, because a technical person helped them get unblocked, they felt they owed something to people.

[00:17:46] Martin Gontovnikas: And then when they were like, hey, now that I helped you, do you wanna talk to sales? They always said yes. There was a good link to get enterprise people to talk to sales because they felt they owed us because we helped them get unblocked and they were like, okay, sure, we'll meet. Other experiments we've done is hiding the features or showing a warning on it. Hey, the feature we're about to use is complicated. Are you sure?

[00:18:09] Martin Gontovnikas: We also tried to add videos and tutorials and trainings. And I tried this with actually very different type of users. And what was interesting was that to me, I always think of users having this bullshit detector. And some users have more than others. So for example, when we tried this with developers, developers hated it if we hid the option from them. Because they were like, I see it in the docs, I see it in some pictures, other people are using it, why don't they have access?

[00:18:40] Martin Gontovnikas: So we started to get a lot of support tickets, people complaining on twitter and stuff like that. However, when it was more like a warning, it worked because it was like, okay, they are letting me use it and just giving me a warning. However, when we tried this actually same thing with other users which were recruiters, they didn't care that it was hidden. They didn't complain, they weren't looking into the dog to see if they were others, and in that case, actually hiding them worked a lot better.

[00:19:05] Martin Gontovnikas: Because when we did the warning, they weren't even reading the warning. They're like, f*ck it. I'm going to use it anyway. So in those cases, it was different and we actually had to hit it to make it work. So to me, what's interesting was that the action that we needed to took based on the learnings was very different, depending on the person and depending on how much outside research or how much formula as well. They were feeling.

[00:19:28] Andrew Michael: Yeah, it makes a lot of sense, I think, in the context of the personas as well, because the ones who are really actually going to be doing the work are going to be the engineers and the ones perhaps maybe like a PM or somebody wanting just to get things set up for the team are not necessarily ones going into the details and being able to hide that. I could see how that is frustrating as well for engineers, those who are on their personality and especially going to docs and like what the hell's going on. Yeah. Like, what are you guys trying to do? Probably going into console and seeing if there's any flags.

[00:19:56] Martin Gontovnikas: Exactly. But the point was we tried this with Ashby, which is a product that is targeting recruiters. And in that case, actually like they didn't care that there were less options or less things, which to me was fascinating.

[00:20:11] Andrew Michael: Yeah, no, definitely. I think it makes a whole lot of sense in terms of the persona and wanting to get on with things and wanting to move along and definitely like being a lot more sophisticated technically in terms of like the product and that side of things. Very cool. So this metric then as well internally, you figured it out, you're starting to experiment with a couple of different areas. Is this just something then afterwards that goes in perpetuity or is it something that you look back at again over time?

[00:20:26] Andrew Michael: And because as you mentioned, like back then, multi-factor authentication was a lot more complicated. Now it's a lot more ubiquitous and everywhere. Like when do you go back to revisit this to try and understand, okay, like maybe our products at a different stage now and we can actually make things easier for them or what does that look like?

[00:20:53] Andrew Michael: We always revisited every six months. Similarly to the activation retention metrics. Like every company I work with, every six months, we always try to look if the activation metrics and the retention metrics we picked in the past are the same, mostly for two reasons. One is, I always think about this crossing the [inaudible]. And as time goes by, your target market also changes from innovators to early adopters to late adopters. So on that side, it changes. But it also changes from your product on it's easier to implement, and it's easier to do. So with that, we actually check with the multi-factor authentication feature. It actually changed not because it was easier in the product, but more because when we had multi-factor authentication, nobody was using them. But then a lot of people started to talk a lot about security and why it matters. There were gaming companies that started League of Legends, started to give you like free things if you used multi-factor. Some other companies started to enforce it.

[00:21:55] Martin Gontovnikas: So once people started to get what it is, how it works, then it made more sense to use it earlier in that case. However, in other cases, it was more about we made the feature a lot easier to use. And because of that, now you can start using it. But what's crazy about it is that we've actually had sometimes where they were using a feature and it was easy to implement, but as they started to add and do other things before, it became more complicated in the future. So sometimes it changes which are the features that are getting people blocked or not.

[00:22:36] Andrew Michael: It's interesting. It's a concept as well, like that we came up with. So not necessarily features alone, but when we chatted to Eleanor Dorfman from Segment, I talked about this episode a lot. One of the things they realized as well was that they would just allow people to get started, self-serve in the beginning. And then they realized actually that companies that added a tracking plan early on and like actually took the time to figure out what they wanted to track. We're always going to be a lot more successful than the ones that just got on and started using the product.

[00:23:03] Andrew Michael: And I think it's a similar concept to this in the sense that there are certain activities that you actually need to do to get things right. But then also there's another layer on top of that. You're saying as well as that perhaps like these activities, you should store them until you've got these first few things right and happy with them.

[00:23:19] Martin Gontovnikas: I think it depends in some cases on whether your product is more self-service or less, like to compare Versel without Xero. In Versel, 50% of the revenue is self-service. There's a lot of people that will never talk to sales, and they will just implement it and do it. So in those cases, I'm a big believer in anti-retention features. But in the other 50% that maybe talks more to sales, you can find other ways to explain it.

[00:23:45] Martin Gontovnikas: For example, one thing for technical products that I've seen be very successful is having an architecture session, where for big customers, you bring your chief architect or your enterprise architect to talk to them, to better understand how the features flow, how they connect to others and stuff like that. And by doing that, not only is that a differentiation that maybe other products don't do, but also is something that helps them better understand and get to implementation sooner.

[00:24:13] Martin Gontovnikas: That was one thing that to me was fascinating, which is that when you compete, like Vercel, for example, they used to be competing with some of these big companies like Heroku or something like that. And in those cases, maybe a big company can send an architect to talk to you. So having an architect talk to you can also be a big differentiation as well, because they feel that you understand their language, you get them and you actually help them to move forward.

[00:24:41] Andrew Michael: And yeah, definitely, I think you can handle objections and figure things out a lot easier when you're on a call and you can see people's responses. I think this was sort of like one of, and it goes to the sort of like one of the pieces of advice I was given like in the past when pitching startups, like on stage. And somebody sort of said to me this once, like, you should limit the amount of details that you share when you don't have direct interaction with your audience because you can't handle objections effectively.

[00:25:06] Andrew Michael: And I think the same thing applies as well as like what you're saying here too, in terms of like the onboarding experience and like what you're giving to your users is that when they get stuck, you don't see their reaction. You don't understand, like you maybe have recordings that you can go to, but you're not there in that moment. And you can handle objections on a call when you're walking people through and you can see when their faces twitch or they look like a bit confused and which you're not getting those signals just on the self-serve side of things. Makes a lot of sense.

[00:25:31] Martin Gontovnikas: And that's why I think guiding them is very important. One thing that Atlassian used to do that I loved and I implemented in a couple of companies is this concept of next best action. And now you can even do it better with AI, which is based on the things that they've done in the product in the past and what they've configured, what is the next best action for them to use?

[00:25:52] Martin Gontovnikas: So then instead of just using the onboarding, you constantly have something called the next best action where you try to guide them and not use some of these anti-retention features, but also you try to get them to use some of the features that will actually most people at their similar stage need to then become retained.

[00:26:11] Andrew Michael: I like that. How do you go about setting that up? Next best action, what does this look like?

[00:26:16] Martin Gontovnikas: On the next best action, how we set it up basically is with a predictive model. So we've got a predictive model using a markup chain, looking into, okay, we have all of the different cases with markup chain of all of the potential path that they've taken, and success to us is both retention and either self-service pay or a qualified opportunity.

[00:26:40] Martin Gontovnikas: So with those, we compare with Markov chains are more successful. So then at any given point in time, depending on what they've done in the past, there's always a future Markov chain that will make them more useful than anything else. But to be honest, you don't need like a data engineering thing to implement it. Like every time we implement it, we've done it in house with some data scientists, but it's not something that took us that much time.

[00:27:05] Martin Gontovnikas: It probably took maybe a couple of weeks to implement, but then the output has been great because at any given point in time, you can always tell a user what to do next. And it's something that you can use not only in the dashboard, but also sending emails to them on when they're stuck or what they should do next and stuff like that. So I personally think it's a worthy investment to nudge them in the right way.

[00:27:26] Andrew Michael: Nice. So you eventually using a Markov model and you're laying out the different steps that users are taking. And then the final end state, whether it's successfully being retained or whether it's like qualified or revenue related, and then being able to sort of understand at which stage each specific user is on the chain and then what would be the next best step to take them to a path of success. Very cool.

[00:27:50] Martin Gontovnikas: To give you an idea how it works is more of like, okay, they've taken these five paths in the past and now we have some people that have done something similar from those, the ones that eventually got to retention which other path have they done next? So we look into all of the potential changes from the past and pick the one that the retained users have done the most next. So it's a probabilistic next steps basically.

[00:28:17] Andrew Michael: Yeah, nice. And it's a lot more contextual then as well to the specific usage. And I can think of it like, so an example for Hotjar, I think Hotjar had like eight different tools at the time when I was there. So they had heat maps, they had surveys, they had recordings and so forth. And I think I genuine working on onboarding, you would try to think, okay, like what should be the actions? But people came to us with different jobs to be done. They had different use cases and somebody who maybe just came for recordings and that was going to be it for them.

[00:28:45] Andrew Michael: They were never going to be interested in surveys or heat maps, let's say, but we would always try and force them through that path. With this, what you're saying is that it's like really understanding, okay, people could just come for recordings and they can be successful and they can become customers and they can stick around. But then it's about suggesting, okay, like you've done recordings, maybe the next step is now integrating events so you can see recordings by events instead of telling them to go off and do surveys, which the general population does. And yeah, I love that.

[00:29:10] Martin Gontovnikas: Exactly. And it's a path that will make more sense for them. The other thing in this exact case that you will do, one of the companies we're working with is Trunk. Trunk has something similar to Hotjar, where they actually have four or five different products. And one thing we do now for the onboarding is we actually look into which blog posts they are coming from or which article or which page did they click sign up from? And we guess by that which product they want to use. So the onboarding is only for that product.

[00:29:40] Martin Gontovnikas: And we only start suggesting, if it makes sense in the future, what to use next. But I think for the multi-product cases, thinking a lot where they come from to automatically pick and make the onboarding and the dashboard easier for what they picked, I think it's a big difference.

[00:29:55] Andrew Michael: Yeah, it makes a lot of sense. I think the only thing that's challenging in that aspect is that it takes time and resources and a lot of development to be able to get to the point where you can have these customized onboarding flows. But yeah, I think like what you've mentioned in terms of the next best action, it feels like a very lightweight way to be able to do this, to get up and running and then, yeah. Very cool. What's one thing that you know today about general attention that you wish you knew when you got started with your career?

[00:30:22] Martin Gontovnikas: I think the main thing that I would love to know when I started my career about churn and retention is that humans are emotional. In the beginning, I always thought that all of the decisions were being rational. So then, if you look at quantitative data, you will have enough. But the reality is that most people are emotional. So if they are pissed off with something or blocked with something, they leave the product. If they don't like how you talk to them, they will leave the product.

[00:30:48] Martin Gontovnikas: But maybe if, I don't know, like MailChimp, you have like a monkey that has a high five and then some confetti, they love it and stay more just because they love looking at that and feeling right with it. So to me, it's mind blowing how emotional we are with some of the things and how trying to drive emotions in somebody with stupid things like a monkey doing high five actually makes a big difference.

[00:31:14] Andrew Michael: Yeah. No, I think that makes a lot of sense. I think that's also one of the big things used to always be pushed is like, it's important to have the what's and the why to have like the qualitative and the quantitative aspect because sometimes you can see something in the data and you get something totally different in reality as well. And data tell you what's happening and people tell you why.

[00:31:32] Martin Gontovnikas: And you also need to use your gut feeling. Like I'm a big believer in most growth people. We always talk about now you need to look at data and do this and data, data this and data that, but data can validate your ideas. But most of the great ideas and great things you can do come from creativity and gut feeling and not from data.

[00:31:49] Andrew Michael: Yeah. Nice. Well, Gonto, I mean, it's been an absolute pleasure chatting today. I've loved learning about the concept of anti-retention metrics and now next best action as well. Is there any sort of final thoughts you want to leave the listeners with before we wrap up, anything that you think you'd want to share or how can they keep up to speed with your work?

[00:32:08] Martin Gontovnikas: The main thing that I would share is spend the time reading about things that are not about the growth or technology and think about how to apply that to technology. That to me is what will give you the edge, the creativity and the uniqueness. If you read and do what everybody does, you'll never do something unique.

[00:32:27] Andrew Michael: Yeah, I love that. And I think that's also something like David used to say a lot at Hotjars, like screw best practices in the sense because like the best way to get mediocre results is follow best practices. And one of the things I'm actually trying to do on the show now as well is get more people from outside in different industries and different spaces to share their perspectives of how they see retaining customers outside the software environment. Cause I think there's a lot of interesting things to be learned.

[00:32:54] Andrew Michael: And yeah, one person I really want to get on is coach Bennett from Nike running up. I think like also just understanding different psychology and the way retention is dealt with in different scenarios, whether it's like the training or it's in customer service.

[00:33:08] Martin Gontovnikas: I agree. And that's where you need this from. So yeah, we'll have to listen to some of those episodes.

[00:33:13] Andrew Michael: Amazing. Well, thank you so much for joining and I wish you best of luck now going into the New Year and look forward to see where HyperGrowth Partners goes from here.

[00:33:20] Martin Gontovnikas: Thank you. And last thing I'd say is if anybody wants to contact me, feel free to DM me on Twitter @mgonto and if you're looking for some advisorship help or GTM help, contact us at HyperGrowth Partners. Happy New Year and thank you for inviting me.

[00:33:41] Andrew Michael: Cheers.

[00:33:42] Andrew Michael: And that's a wrap for the show today with me, Andrew Michael. I really hope you enjoyed it and you were able to pull out something valuable for your business. To keep up to date with and be notified about new episodes, blog posts and more, subscribe to our mailing list by visiting Also don't forget to subscribe to our show on iTunes, Google Play or wherever you listen to your podcasts. If you have any feedback, good or bad, I would love to hear from you.


Martin Gonto
Martin Gonto

The show

My name is Andrew Michael and I started CHURN.FM, as I was tired of hearing stories about some magical silver bullet that solved churn for company X.

In this podcast, you will hear from founders and subscription economy pros working in product, marketing, customer success, support, and operations roles across different stages of company growth, who are taking a systematic approach to increase retention and engagement within their organizations.


Listen To Next