Boosting Marketplace Retention Through Engagement and Quality Management at Scale AI

Ruslan Nazarenko

|

Head of Growth

of

Scale AI
EP
270
Ruslan Nazarenko
Ruslan Nazarenko

Episode Summary

Today on the show, we have Ruslan Nazarenko, the Head of Growth at Scale AI.

In this episode, Ruslan shares his approach to boosting retention in marketplace models through strategic engagement and quality management. He dives into Scale AI's unique model for retaining skilled gig workers, emphasizing how tailored experiences can overcome common marketplace retention challenges.

We then explore the role of dormancy and reactivation in maintaining a resilient workforce, and how Scale AI balances high-quality standards with user engagement to create a robust marketplace ecosystem.

We wrapped up by exploring the critical role of company culture in driving sustainable growth, the value of setting ambitious goals, and Ruslan’s insights into building a resilient, motivated team that drives growth.

Mentioned Resources

Highlights

Time

Introduction to Ruslan Nazarenko and Scale AI00:01:02
Lessons from Quokka: Retention Through Engagement00:04:03
Marketplace Retention Challenges at Scale AI00:07:19
Strategies for Dormancy and Reactivation00:10:44
Balancing Quality with Engagement in Marketplaces00:16:41
Importance of Iteration and Customization in Growth00:20:18
The Role of Culture in Driving Growth00:29:58
Key Takeaways on Building a Resilient Marketplace Workforce00:38:01

Transcription

[00:00:00] Ruslan Nazarenko: What we also learned is that we have obviously a much higher chance to retain people if we just keep them engaged. Once they stop being engaged, that's when we have quite a high chance we're going to lose them. So the question is, okay, how do you keep them engaged at the end of the day?

[00:00:00] Andrew Michael: This is Churn.FM, 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.

[00:00:38] 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 bootstrapped, profitable and growing.

[00:00:51] 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.

[00:01:02] Andrew Michael: Hey, Ruslan, welcome to the show.

[00:01:04] Ruslan Nazarenko: Thank you. I'm excited to be here.

[00:01:07] Andrew Michael: It's great to have you. For the listeners, Ruslan is the head of growth at Scale AI, the data engine that powers the most advanced AI models. He's also an angel investor and advisor to companies like Substack, Respeecher and Reface. So my first question for you, Ruslan is, what is your biggest learning that came out of Quokka?

[00:01:25] Ruslan Nazarenko: Oh, Quokka, dude, this is great. So that was one of my startups. I had a lot of them. Okay. And with Quokka, the idea was that if you try to send an email campaign, like what's your open rate? Is it like 30%, like 20% average, right? And I was thinking back in the day that, Hey, it means that like 70% of people, they will not even get your message. And what can you do about that? And obviously, you can experiment with the titles and stuff like that, but it still will not get you close to 100%.

[00:02:00] Ruslan Nazarenko: So what we did is we took the audience of people who have not opened your email and we created a custom audience for you on Facebook or Google or whatever. And so then you can pretty much deliver the same message across other channels. And we did get a real attraction. We had quite a few big names. But what we realized is that this thing is so niche that there's no way you can build a scalable business out of it. And then I got a job at Bird and we ended up just shutting down the company, which was sad.

[00:02:39] Andrew Michael: Yeah. I definitely see that, how it could be... It seems almost just like a feature of a product as opposed to a product.

[00:02:46] Ruslan Nazarenko: Exactly. That's actually what happened. That's what happened. A lot of people... So we partnered up with a lot of CRM tools out there, and those tools that we did not partner up, they built it as a feature internally. And we're like, damn it.

[00:03:03] Andrew Michael: Yeah. And I think also, I mean, it's something I've always wondered and thought about as using as a channel for engagement and reactivation with existing customer base as well. We use ads and stuff for acquisition, but very rarely do we think about it in the context of retention and how we can actually bring people back into the app. And I think we rely so much on this lifecycle marketing and emails, but there are other channels that we can actually be using to bring them back into the app. So I think it would have been... Did you see many people using it for that use case or was it mostly for acquisition and sales?

[00:03:37] Ruslan Nazarenko: No, it was actually the engagement and retention component. So I think this whole product was born out of the anecdotal experience that I had because I was working with a B2B SaaS company back in the day and that's literally what I did manually. I created the audience, I decided to give it a shot just as an experiment and it turned out to be great. I don't remember the exact number, but we got a huge spike in revenue.

[00:04:03] Ruslan Nazarenko: And I'm like, damn, that thing works. And I was just like, trying to find a friend because like we did not have a lot of internal resources. So I was trying to like, reach out to my friends and I was like, hey dude, so this is what I did. Like, can you build me like some kind of script or something? And he was like, well, I definitely can. But then also like, why don't we try to build a product out of it? I'm like, oh, I don't know. Well, I mean, let's try, I guess. And then yeah, ended up trying.

[00:04:31] Andrew Michael: Scratching your own itch. Yeah, I had the most of the best products you speak to, like, end up coming out of something like this and they evolve and figure out ways to grow from there. But it's always an amazing experience, I think, that you come away with and a lot of learnings.

[00:04:46] Ruslan Nazarenko: Yeah, I was about to say that that was a good experience. But, you know, it did not turn out to be a good product. So there's that.

[00:04:55] Andrew Michael: I think it's incredibly hard to go from that zero to one for the easy, to build something, but then to get traction and to actually start scaling and to get product market fits. It's such an elusive thing that you just need to keep iterating at. Before you joined, we were chatting a little bit and you said you were out in Portugal, you were about to open up a bakery and then Scale AI found you. Why a bakery? And what were you thinking?

[00:05:22] Ruslan Nazarenko: This is like one of the most popular stories in my life right now. But what was happening is, I moved to the States for the first time in 2020. And then around, like 2022, I moved to Portugal. And I ended up living in a small tiny village in the middle of nowhere. So no restaurants, nothing like that. And at that moment, I didn't know how to cook. Literally, I can boil the eggs. This is my, like the maximum, the top of it.

[00:05:54] Ruslan Nazarenko: So I basically had to learn because I mean, I'm in the village. I'm a foodie. I love good foods. So I had to learn how to cook. And I started with… Some reason I started with baking. I'm like, okay, I'm going to try to bake something. And long story short, I went to Paris to one of their schools. I learned how to bake and I was like, okay, maybe I want to open up a bakery because I'm kind of, like, sick of the tech a bit. But yeah, Scale is probably [cross talk].

[00:06:21] Andrew Michael: So you took it very seriously. Yeah. Nice. I mean, I think in some ways it is, like, a very close to growth because there's a lot of experimentation. Like once you get a good recipe, like you'd need to hang onto it, but there's always things you can tweak to perfect it and you're constantly iterating on it and like I did the same thing with pizza during COVID. So like I went all in and like, yeah. So I like my room, like when I make pizza now, like I have a certain degrees and humidity levels and like, I want everything to be perfect, which is really stupid. There's no need for it, but like, I enjoy geeking out on it. So I definitely see the obsession.

[00:06:53] Ruslan Nazarenko: This is the same. I literally had an experiment tracker on my breath. Oh man, it's crazy. But yeah, it's what it is.

[00:07:06] Andrew Michael: But you gave up on that dream, maybe hopefully temporarily, and at some point you'll come back to it. You've joined Scale AI now. Can you just explain a little bit about Scale AI, what it is that they do and what you do within the context of it?

[00:07:19] Ruslan Nazarenko: Yeah, I think the easiest way to explain Scale to people is that all those LLMs, they need data. And the better the data is, the better the model turns out to be. And so what we do is we provide companies with really high quality data. Maybe I'm obviously biased, but I would say it's the best data that you can potentially buy in the market. And if you think about it, this is essentially like a marketplace, right?

[00:07:49] Ruslan Nazarenko: Because you have like, whose companies on one side, this is your demands, they're telling you like, you know, oh, we need like 500 hours here, we need like 5,000 tasks there. And then on the other side, you need to like, find people who will be able to kind of label that data. And so I'm on the supply side. I'm actually on the side where, like we have gig workers who come to the platform, they work on projects, they label data, and I'm responsible for just, like retaining them essentially and making sure that we utilize the capacity that we have.

[00:08:26] Andrew Michael: Interesting. Yeah, because I had a sort of a decent idea of what Scale AI, but I never realized there was a marketplace on the back end. This is how things operated.

[00:08:36] Ruslan Nazarenko: Yeah, it's kind of like a managed marketplace, right? Because we don't have, like a sales server, at least not at the moment. So yeah, I tend to not show it as, marketplace.

[00:08:44] Andrew Michael: And then you have like on the client side, it's more like B2B sales driven or like how do you onboard new customers to the platform? Because I guess there's a lot, like you need to manage the inventory that you do have and on the gig workers perspective, like to be able to support the work that needs to get done as well.

[00:09:03] Ruslan Nazarenko: Yeah. So on the demand side, this is B2B sales. And then because like we're talking about really huge checks. It's like, not even like tens of thousands per month. It's much, much more.

[00:09:17] Andrew Michael: Yeah. And what is, like the day to day in your role then look like now if you're focused on retention for the supply side, like what is the work involving for you?

[00:09:30] Ruslan Nazarenko: Yeah, we can maybe take an example, right? So there's a new project and we know, like the plan, like the model for that project. We know that they want to do, like there are five something hours, like this week is going to be like 10,000 next week, like 15 the week after it's around. And so then you just sit down and you essentially build a gross model. Because you're like, okay, how many people, we have in the system? And within those people, do they work on other projects? Or are they retained? Are they churned? Are they dormant? What's going on there?

[00:10:05] Ruslan Nazarenko: And then you can figure out, okay, we have that many capacity available. We might have 2x more if we are able to retain or resurrect those people. And then the gap, we potentially need to close that with tofu, so just like new user acquisition. And then it's like, again, they're operational, so you just sit down, you're okay. In my case, I want to resurrect people. So I need to figure out who are the people that I want to resurrect. So just like you write a bunch of queries, you're trying to understand their past performance, where they are now, what they have done in the past, and then usually know that they have had some kind of problem.

[00:10:44] Ruslan Nazarenko: Maybe the previous project was not, ideal fit. And so for that group of people, you would just tell them that, hey, now we have a new project that is way better fit for your skills. So if you're still interested, feel free to give it a shot, try it out. That's the first phase. You launch the project, then obviously you have all the drop-offs throughout the final. Some people, they don't even get started on the project. Some did start on the project, but they just did a couple of tasks. So that's again the question line, how do you retain them? What do you do?

[00:11:18] Ruslan Nazarenko: And then we are doing our best again to ensure the quality, which means that some people are getting filtered out because we don't think that they are high quality enough. And so from the retention side, then you just take that forward and you potentially just talk to them trying to figure out like, hey, you seem to be a PhD in biology.

[00:11:39] Ruslan Nazarenko: Like how do you fail the screening course? What do you mean? And it's obviously usually not because they don't know biology good enough, because they have PhD in biology, right? So it's us, did something wrong. And then the question becomes like, what do we do with those people? How do we reverse them essentially?

[00:11:58] Andrew Michael: Interesting. So it's like resource management and a whole nother level where you're working with independent third party, like contractors to some degree. And then you need to manage like, not only matching like individuals to work, but actually with work that actually meets their needs and their skills and activities as well. And then so it sounds like most of these people then would be like highly skilled professionals that would be doing the service. Do you have a mix of–

[00:12:24] Ruslan Nazarenko: We have a mix. Yeah, we have a mix. Like typically you can think about it as like four main buckets, right? You have like specialists, those are like your PhDs and stuff. You have languages, you have coders, and then you have just like generalists essentially. So, and then, yeah, based on the project, you can use the mix of them. You can go with one pool, so it can enhance.

[00:12:46] Andrew Michael: Okay. So in practical terms, I come to, let's say I want to build a new app to be able to study and identify MRI scans and be able to give prognosis or something like that, or labeling MRI scans. You would then have doctors who would take a look at each one of those scans and then label them, and then that's what essentially the clients would be paying for to be able to train their models on that data. Is that fair?

[00:13:11] Ruslan Nazarenko: Yeah, that's fair. There's a bit of a back and forth going on, right? Because especially as a startup, you can think that, oh, I need this thing. And then you get this thing and you're like, oh, well, actually, we need some more data or we need a different type of data. So now there's this, run the pilot, run the projects, redo the pilot, redo the project type of collaboration happening.

[00:13:36] Andrew Michael: So it sounds like then like your interactions and the work that you're doing currently, today is like, very much like one to one almost in a way, in less of like one too many. Like how are you scaling, like your efforts when it comes to growth? Obviously like I think in the resurrection side, I think that makes sense where you can run campaigns, but then on the other end, on like the quality aspect and maintaining that, like how do you Scale this program that you're working on?

[00:14:02] Ruslan Nazarenko: Yeah, this is actually where it becomes really tricky because it's… At the end of the day, it's not one-to-one, right? Because we're still talking about like hundreds or thousands or like tens of thousands of laborers. And the issue or not the issue, the interesting part here is that every single project is like, super unique. And so it's really, really hard to build something that is like, easily scalable because like every single time, is every single new project, you have to like, tweak your queries, tweak the ways to like do stuff and see how much of that you can kind of try to productize at least to some degree.

[00:14:37] Ruslan Nazarenko: And so this one-to-one, this is usually the very beginning, right? Because in the quantitative data, you see that you have a drop-off. To figure out why the drop-off exists and what do you need to do, you go and you talk to people. You can talk to people, you can survey the people, but somehow you need to get an idea of what is going on there. And that's where this one-to-one interaction happens.

[00:15:01] Ruslan Nazarenko: But then once you get it and you maybe try to solve it, like this one specific customer and you see it clicks, they just magically understood how to train the model. Then you're like, okay, that works. Let me try to tell the same thing maybe through the email campaign or maybe through a webinar, to the entire user base and see what happens. So this is very fast iterations I would say.

[00:15:33] Andrew Michael: And are you doing that then on, like a campaign by campaign basis? Because like each campaign would then be unique and what needs to be done. So, like, customer comes in, you start the study, you start seeing, like things maybe going wrong in a specific area. You're getting qualitative feedback, you're learning, and then you're sending out campaigns to everybody else who's about to go ahead and take that or let us start the labeling or the data.

[00:15:54] Ruslan Nazarenko: Yeah. And, and then kind of like over time, you still realize that, yeah, every single project is different, but then there's some kind of like a shared taxonomy between all of them, right? And you like, start to bucket the projects, you can start bucketing the problems and this is how you ended up in this product or productized experience to solve that. So I would say that we kind of switch from this very manual approach now to more thinking about on a cohort or cohort basis, right?

[00:16:23] Ruslan Nazarenko: Because once you understand that, okay, we are not doing a great job in activation, you change the project, you change the onboarding experience and then the next cohort that comes, it's like slightly different from the one that you have seen like the last three weeks, let's say. And so your approach to that core is also different.

[00:16:41] Andrew Michael: Yeah. We're actually chatting before the show as well about Adam Fishman, because I had him on the show like a couple of weeks ago. And our discussion now is reminding me a lot of similarities and towards, like the discussion there, where there's a lot of, like seemingly things that are maybe outside of your control that could cause other like the customer churn or on the other end, like the freelancer's churning.

[00:17:04] Andrew Michael: And there's probably a lot of things that you can do to put in place or tools that you can help them with when it comes to training or when it comes to setting up the projects or guidance and all of this that will help streamline these processes. And it's very much a qualitative game. So I'm keen to hear a little bit about diving into those qualitative interviews then and what are some of your go-to questions or methodologies that you employ to actually try to pull out what's going wrong in these studies?

[00:17:31] Ruslan Nazarenko: Yeah, I feel like it's really, really simple, at least in our case, right? Because when I see the drop off, like I basically see how many people dropped in a step, right? And I know that on average like, it takes two days, let's say, to go, like, first step to the second one. So basically I take everybody who have not moved to the second step within like three days, and it's just like the first iteration, I probably would just send them, email, like literally asking like, hey, what's wrong? I see you're stuck. I see you're not doing the training. I see you're not doing the work.

[00:18:04] Ruslan Nazarenko: It depends on the stage of the funnel, right? But tell me what's wrong and you can just reply to the email. And so the next morning, I will wake up to hundreds of replies in my inbox and I would just manually go to a bunch of them trying to see if there's a common theme that people mentioned. And if they do, sometimes it can be enough. I'm like, oh my gosh, okay, we don't do the benchmarks correctly. The benchmark is like when we try to assess the quality of people on the internal data, rather than on customer-facing data. So that can be it.

[00:18:37] Ruslan Nazarenko: Or in other cases, I still don't have a really good idea, but I kind of understand where it's going, so I can come up with kind of like a survey with more closed-ended questions and send it to a bigger user base to see if I can fish for some kind of insights in this way. So yeah, that's really pretty straightforward.

[00:19:01] Andrew Michael: It's an interesting one as well though, then ‘cause like, the focus is then on the people that aren't successful and less on the people that are successful as well. And I think like, this is something like I've talked, on the show before. I loved was like a lesson from [David Dominan] on, like, how to optimize websites. So it's normally like you should focus on those that have just been successful and ask them what nearly stopped them from completing their action.

[00:19:25] Andrew Michael: Because then you're getting like the 50% who maybe had been a little bit confused, but they actually went through and those that are always going to go through, go through and then you can then figure out like what you can do for that like portion of users that were like the other 50% that are dropping as opposed to maybe those that weren't a good fit. And I wonder if that's something that you thought about applying or tested.

[00:19:43] Ruslan Nazarenko: Yeah, so we definitely thought about it. The thing here is people are coming here to earn the money, right? So we have way more leverage compared to like a lot of other marketplace out there like you’re coming to, like there are marketplace like Etsy to maybe like, buy stuff. So you're actually spending money, right? Which in our case is that people are coming to earn money. So they have, like, much higher motivation to like, go through all the pitfalls, like, figure everything out, forgive us for like all the shitty UX that we might have because they want the money.

[00:20:18] Ruslan Nazarenko: And so in this case, the main blocker for people to earn the money right now, it's not their motivation. It's literally us. It's like, oh, we configured the projects not in the right way, or oh, we put the wrong people to the wrong project. So you see, this is not about you as a site or a gig worker. You're doing something wrong. It's likely us doing something wrong. And so this is why we're so much more focused on people who went dormant compared to people who are just staying on the platform.

[00:20:54] Ruslan Nazarenko: Because we know why those people are staying on the platform. And we know that they're well, they make money, and some of them are excited about the AGI and stuff. So this is why. Yeah, what about everybody else who wants to work, but they just cannot work because we did something wrong.

[00:21:11] Andrew Michael: Yeah. So you said then you do a lot of time on resurrection and trying to bring people back to the platform. What are some of the things that you've been trying and you've seen good success with? And I think it's a little bit unfair though, in this case, because as you say, and pointed out is that these people are actually getting paid by you and they're less about, like coming back to use my products. But I'm keen to hear if you've used any sort of tactics that you've seen work in the past and not being, applying at Scale AI.

[00:21:36] Ruslan Nazarenko: Yeah. I think, like the first breakthrough for me personally was when we stopped thinking about, like churned people really. Before, when I just joined, we were like, well, we have some people who do work and we have people who churned. And I'm like, is that the right definition actually? Or is there something more? And this is where I can reword it to the reforged model where I'm like, okay, maybe we have truly churned people as is. We don't have a high chance to reactivate them. And then we have dormant people. And dormant people, we have a higher chance.

[00:22:13] Ruslan Nazarenko: And so if you're going dormant, why are you going dormant? Is that because you have work available, you just decided not to work? And if so, then what kind of reasons might lead you to this decision? Or it's more so that you want to work, you are coming back, you don't do any production work. And then the question becomes, okay, again, what kind of blockers do we put on your user journey that prevents you from working actually?

[00:22:42] Ruslan Nazarenko: And this is like a really tricky balance because there are, like some good people, obviously good people, right? Who are trying to do their best. There are some people who are trying to achieve the system. They just want to get money. They don't want to do, like, any good work. Right.

[00:22:55] Ruslan Nazarenko: And then there's like a whole world of like people with different intention, different intuition that are coming to the platform. And somehow you need to like, maintain the balance between reactivating the right people in the right way like trying to dodge the fraudsters and cheaters and scammers. And that's what makes it, like, even more complicated, to be honest.

[00:23:18] Andrew Michael: I was just saying, I was thinking, I wonder how many people are training models to answer questions for Scale and to be able to go through the... Yeah.

[00:23:26] Ruslan Nazarenko: That's a lot of people. That's a lot of people. Yeah. We do invest quite some resources, like trying to build this whole, like anti-cheat systems. It's just like, yeah, exclude people who are trying to like, use LLMs, to build LLMs.

[00:23:44] Andrew Michael: It reminds me a lot of like panel providers as well, to a large degree, who are filling in surveys and having worked with that quite a bit for pricing and packaging research, a lot of work needs to go into making sure that you get good, clean data at the end, even in the end, which I sort of think it's different. Like are you still paying them as well? I'm not sure what you pay at Scale versus like a Panel typically does, but it feels like very similar jobs, like labeling versus answering specific questions on a survey and yeah. How are you dealing then, with the quality issues?

[00:24:20] Ruslan Nazarenko: The quality issues is we do have a set of metrics, right? That they all try to describe the same system, the quality system. We don't have a single view right now. Like, you know, it's not like you can pinpoint me to this person and say, like, oh, are they doing a good job? And I'm like, yes, they do. Or No, they don't. It's usually way more nuanced than that. It's like, oh, well, that person have done a great job at this type of project, but not this type of project. And they did a great job with this skill, but not really this skill.

[00:24:53] Ruslan Nazarenko: So it depends on your needs. They might or might not be a good fit. Which again, when we talk, this one-to-one thing, that's easy. Right? But then if you're coming to me and you're telling me like, oh, I need like 20,000 good people, then like, what do we do? And in case of Scale, we actually do like a few things. We, throughout the life cycle, we measure the quality. Like we measure the quality when you just like, join the platform, you have not done any, like production work. We'll try to get, like, a quality rate on you.

[00:25:28] Ruslan Nazarenko: Then during the training sessions, we're gonna, like, ask you a bunch of questions to also like, try to get a… an idea like how good you will perform. And then we constantly measure your performance as you go through the projects. And again, this is why we have, like this set of different metrics that describe, like different parts of the user journey. And then depends on your needs, we can be like, really strict and we can try to slice to like the best people possible.

[00:25:56] Ruslan Nazarenko: Or if you need, like way more people, we can lower a bit the barrier like see what happens and if it's okay, then okay, that's fine. If it's not okay, then we got to, like probably implement additional training and additional screening.

[00:26:11] Andrew Michael: What's been your biggest surprise then since joining Scale when it comes to working on, like retention for the business?

[00:26:18] Ruslan Nazarenko: I would say we do way more manual work than I was expecting. And I didn't know that you can be at such a Scale. Pun intended.

[00:26:31] Andrew Michael: Pun intended, yeah.

[00:26:35] Ruslan Nazarenko: And operating this like, manual way. And when I just joined, I was frustrated. I was like, why are you just doing it? Why don't you have, like a triggered email? Why don't you have, like all these campaigns? Like, you know, the basic stuff. And I was sent in the trenches basically, because they were like, well, you go figure it out yourself. Like you will find out soon. And so yeah, I basically fuck around to find out that it's really because of all this customization. And in like–

[00:27:08] Andrew Michael: It was actually one of the questions running through my mind as well while you were explaining the setup was, why not just set up an automation? It seems obvious.

[00:27:16] Ruslan Nazarenko: Exactly. Yeah, it seems obvious. There's a lot of things that seem really obvious, but they're not here yet, at least. And it's because in 100% of all the use cases, the projects and stuff, there's maybe only like 3% of them are like really common and that you can build the automation. Because, let's maybe take a look at the example here.

[00:27:39] Ruslan Nazarenko: Say you want to reactivate people in the project. You want to take, like everybody who have done any work on the project but not done any work in the last seven days and you want to, like send them an email saying that, hey, there's like tasks available, right? Pretty straightforward. But then are the tasks available? What if there are no tasks available?

[00:28:00] Ruslan Nazarenko: But we do know that there's going to be more tasks uploaded to the system tomorrow, let's say. And the customer expects us to deliver those tasks in 24 hours. So your reactivation window is actually, it's not days or weeks, it's hours. So somehow you need to let people know that there is going to be the work available so they can plan their schedule around your work. But then also, customer might take longer.

[00:28:26] Ruslan Nazarenko: And if I tell you, like, hey, Andrew, there's going to be work tomorrow available. You're like, okay, fine. I'm going to change my schedule. And then it's tomorrow already, but there's no work because the customer said that, oh, we need like an extra 12 hours to redo something. And here we go. You're sitting there doing nothing. I'm sitting here. I reactivated you technically, but I'm not getting any work because there's no work available. And so that's kind of what makes it so, so complicated.

[00:28:55] Ruslan Nazarenko: And the beauty of it is that you have to rely on a lot of proxy metrics and a lot of signals to actually trigger the campaign. Because you need to maybe analyze the customer sentiment for the last project you did. And you're like, okay, they're really happy. We know that they want to Scale really fast. So they likely will upload the task when they said they will.

[00:29:20] Ruslan Nazarenko: And so that means that for this specific project, we actually can send the early heads up to people and let them know that there's going to be work available. And even in some cases, we can let them know how much work will be available. So that's kind of hard. And how do you productize that? That's the big question mark.

[00:29:40] Andrew Michael: Very interesting challenge, I think, to work on. I see we're almost up on time, so I want to make sure I ask you a couple of questions I ask every guest. The first question is, what's one thing that you know today about churn and retention that you wish you knew when you got started with your career?

[00:29:58] Ruslan Nazarenko: No, that's a good one. I would say it's this kind of segmentation that I mentioned, right? It was like voluntary and involuntary dormant people. But then also, like what I usually try to do now at Scale and say, I did a third female and much of other companies is, I'm trying to go into like theory specific problems that people have. And then the tricky part is, again, how do you track any given problem at Scale? How do you know that, okay, there are this many people who face the problem X, that many people who face the problem Y?

[00:30:34] Ruslan Nazarenko: And then you try to work on... Rather than working with all voluntary dormant, all involuntary dormant, you're trying to work with very specific subsides of issues for very specific subsides of customers. And kind of like usually requires a lot of, like, being [not] work. And this is still like something that I didn't find like a better approach to be honest. It's like literally, I'm repeating myself, but this is like when you almost go, like one to one before you can go one to many. And how do you build the system there to attract that one to many relationship?

[00:31:07] Andrew Michael: Yeah, no, absolutely. I think it's like, I think about this as well in the context of, like activation and onboarding. Like you're far better off just like, I think we spoke to Superhuman as well. We had Gaurav recently, but just like in the beginning, like doing those one-to-one conversations, figuring out, like what are those main pain points and then finding ways to productize and scale things as opposed to trying to, like serve everybody with a generic solution that's ultimately like–

[00:31:31] Ruslan Nazarenko: Yeah. I think actually I might have like a perfect example here, which happened at Bird, that was not retention, that was activation. Well, we all know. Thanks to Brian, that's activation and the equation as well. Yeah, so we kind of figured that there is a difference between people who write often versus people who don't. Because the distribution of users back in the day was that we had a lot of people doing just one write. We had a lot of people doing like five or more writes and we didn't have a lot of people in between, which is, like, weird. This is not really a normal distribution, right?

[00:32:08] Ruslan Nazarenko: And what we did is we just literally went on the street because we kind of know where scooters are, right? And we were staying next to the Nest. It was called, Nest where you have like a bunch of scooters nearby. And every single person who come to the scooter, we ask them like, hey, why are you doing it right now? What's going on? Why? Like, why?

[00:32:31] Ruslan Nazarenko: And they, we got like a variety of responses. People would be saying, like, oh, I'm gonna meet my friends or oh, I'm gonna like, I'm going to a doctor or I'm going elsewhere, like, you know, they started naming these use cases. And when we came back to the office, we cross-checked, that their response was like, actually what happens with their accounts.

[00:32:52] Ruslan Nazarenko: And what we noticed is that people who wrote like one time, two times top, they don't know about the use cases. They're like, oh, you know, I'm going to my friend, that's about it. Or, oh, I'm just gonna try it out. That’s about it. They never will come back again versus people who are like, oh, I can take the scooter to go to the doctor, to the bar, to the bank, to the restaurant, to meet my friends, to have fun with my friends. Those people, they do write a lot.

[00:33:19] Ruslan Nazarenko: And so there's this kind of mental difference between people in that case, between people who really did get activated versus people who have not come to the activation stage, which is like understanding the use case. And this is something that I keep seeing across the variety of products and companies is that it's really unproductive to focus on just one side of the equation.

[00:33:45] Ruslan Nazarenko: You probably want to slice people based on like, okay, you have the segment of people who are coming back, ask them why they're coming back. You have the slice of people who are not coming back. Ask them why, the mental gap, that's the gap that you want to try to fill. And that's what can potentially help you with your motivation, retention, activation, whatever you want to call it. Problem.

[00:34:07] Andrew Michael: Very nice.

[00:34:07] Ruslan Nazarenko: And again, this is something I learned, I guess, way too late. This is not about just one side of the coin. You need to research both sides.

[00:34:16] Andrew Michael: Interesting. Yeah, I think, absolutely. I mean, on the show, when it comes to biggest influences on churn and retention, activation is one of the top most discussed topics on the show by far. Like top three, maybe even top topics.

[00:34:30] Ruslan Nazarenko: Oh yeah. And I was about to say that I can see why that kind of totally makes sense because what we also learned is that we have obviously much higher chance to have, like retain people if we just keep them engaged. Once they stop being engaged, that's when we like, have quite a high chance we're going to lose them. So the question is, okay, like how do you keep them engaged at the end of the day?

[00:34:55] Andrew Michael: Absolutely. Nice. So next thing then as well, obviously. You're an advisor. You speak to a lot of different startups as well when it comes to growth. What's one thing that you wish more people would ask you, but they don't?

[00:35:12] Ruslan Nazarenko: Ooh, that is a deep one. I think the answer from my side at least, it's how to build the right culture. Because when I started the growth, I did believe that this is about the tools that you use. And you go to like a new fancy tool, you gotta, like find a better way to, like analyze your retention and that what will like make the company grow. I was wrong, obviously.

[00:35:44] Ruslan Nazarenko: And so then I was like, okay, maybe if it's not about one tool, maybe it's about like a bunch of them. Maybe it's about frameworks, right? And if I were to use the right framework, that will surely make my company grow. Also, it didn't happen, obviously. But then you still have Uber and Lyft, right? You have Bird and LIME, you have Scale, and I'm not going to name the other companies, but there's a lot of other companies, obviously.

[00:36:11] Ruslan Nazarenko: Why is that? Why one company keeps outgrowing their competitors? And what I came to realize recently, that the answer is the culture. And it is the culture of usually, usually of unachievable goals. Because if you look, take a look at early, like Uber or Scale, like, you know, our KPIs are quite aggressive to say the least. And we all know that, yeah, we're not going to hit them. This is fine. Let's try to get as close as possible.

[00:36:40] Ruslan Nazarenko: So it was the culture of unachievable goals. It's the culture of high accountability. Like, you can adjust, sit back and kind of like try to fly under the radar, right? You will get asked questions and those questions are very specific. In case of, like Scale, it's well, how many hours did you deliver? In case of Bird, it was like, well, how many rides did you get us? You just, you cannot bullshit in this case. You have to save the number.

[00:37:07] Ruslan Nazarenko: And there are, like quite a few other things that I'm thinking about, like right now in terms of the culture. But I think that this is actually the main reason why, like what makes other like, all things being equal, companies, one of them will grow faster. That's the answer, culture.

[00:37:23] Andrew Michael: Yeah. No, that definitely resonates. I think like sort of having accountability, having sort of urgency and then having big goals that you go after is like critical. It's a critical mix. I think that's what, sort of, like what I felt we had at Hotjar as well. I think it was like the first time in my life where I joined a company and I felt sort of like imposter syndrome. I was like, why am I here?

[00:37:43] Andrew Michael: Like everybody else is so amazing and everybody else is like, so on it all the time and into this. And like, I think like when you're in that environment, it raises everybody around you to be at another level. And it definitely was an incredible experience for me. I think it would be very hard to replicate again or find somewhere else.

[00:38:01] Ruslan Nazarenko: Which is again, how do you even have all those people in one place? Because we all know a lot of companies where you have a few people, they're really good. And then you have the rest of them who are just chilling. And that's about it. So how do you build that culture? I think this is the question that more people need to ask themselves.

[00:38:23] Andrew Michael: You need to be extremely deliberate about it.

[00:38:25] Ruslan Nazarenko: Oh, yeah.

[00:38:26] Andrew Michael: Getting through the experience out there and the processes in place to make it happen. Ruslan, it was an absolute pleasure chatting today. Is there any final thoughts you want to leave the listeners with? Any things you want to mention of how they can keep up to speed with your work or keep in touch?

[00:38:42] Ruslan Nazarenko: Yeah, well, I mean, keep in touch. The LinkedIn would be the best place. I did start my Substack. I'm not doing it actively just because we work a lot. I don't have, really, much of a time to do that. So LinkedIn will do it. I'm thinking about the last, maybe like a couple of thoughts. But yeah, I think we call it as pretty much all of it was the culture. I would maybe say that scrappiness, like just be fast and be scrappy. That's also something that I'm seeing, like [across] other companies, worked really well in the past.

[00:39:16] Ruslan Nazarenko: This is something that works at Scale really good. And I think like especially, was when you're just getting started, was like retention and churn, you have, like so many ideas that it probably makes no sense to try to like, productize that. And you probably want to be, like, as scrappy as possible to run these first experiments and then like to fortify your positions after.

[00:39:35] Andrew Michael: Yeah, definitely. That's, that sense of scrappy is I think that the problem is, like typically as companies grow, they start to lose that and they feel that they need to outgrow it, but I think that's a weakness in its own right. I think you need to be maintaining at least specifically in growth, that ability to be scrappy and to be able to move fast all throughout the journey of the company. It's not just a specific stage or only for startups.

[00:40:01] Ruslan Nazarenko: Yeah. I actually think, since you just mentioned that, this is also something that makes Scale culture unique, right? Because as a growth person, you're not optimizing the third email in a sequence. It's basically like we're literally operating this model where we ask ourselves, what is the biggest blocker? And say, if the biggest blocker is support, then it's a heavy growth. You just go to support and you fix the support function.

[00:40:26] Ruslan Nazarenko: You don't build the whole thing from the ground, but you fix some of the problems that blocks you and then you come back. And so I just wish more people would do that in this way. Don't try to limit yourself. It's like, oh, I'm only doing like retention because you have like a lot of drivers for that metric. And you need to just like, take all the levers that you have and like pull them and see what happens.

[00:40:50] Andrew Michael: Amazing. Yeah, well, Ruslan, thank you so much for joining. I really, really appreciate it. It's been great getting to learn from you on the show and I wish you best of luck now in figuring out these exciting challenges you've got ahead of you at Scale.

[00:41:02] Ruslan Nazarenko: Thank you, thank you. I'm excited. I hope we can talk, in maybe like a year, two years, and I will have better answers for you. Like what worked for us really well.

[00:41:10] Andrew Michael: Absolutely. It's always good to have a catch up and see a fast forward a year, what's happened and the learnings in between. So we can schedule that one. But thanks so much for joining and wish you best of luck now.

[00:41:21] Ruslan Nazarenko: Thank you. Thanks for having me. Bye.

[00:41:24] Andrew Michael: Cheers.

[00:41:26] 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 Churn.FM and be notified about new episodes, blog posts and more, subscribe to our mailing list by visiting Churn.FM. 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 and you can provide your blunt, direct feedback by sending it to Andrew@Churn.FM. Lastly, but most importantly, if you enjoyed this episode, please share it and leave a review as it really helps get the word out and grow the community. Thanks again for listening. See you again next week.

Comments

Ruslan Nazarenko
Ruslan Nazarenko
About

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.

Related

Listen To Next