Lomit is the VP of Growth at IMVU. He is also the best selling author of the Lean Startup series book, Lean AI as well as a public speaker and startup advisor. Lomit has a ton of experience getting companies to and through the rocketship growth phase. Here are just a few of the topics we discussed in this episode:
• Monetization strategies for your high growth startup
• Extracting value out of customer data to help you grow the business
• Identifying the right users to grow LTV (lifetime value)
• How to properly use automation to grow faster
• The inspiration behind his hit book, Lean AI
• Lean AI by Lomit Patel
• Marketing automation tech - Nectar9
• Lomit’s website & blog
Connecting with Lomit
• Connect with Lomit on LinkedIn
Hello, and welcome back to the Customer Conversations podcast. I am the co-host, Sean Boyce. I would like to welcome my guest today who is Lomit Patel. He is the VP of growth at IMVU. Lomit is also a bestselling author, public speaker and startup advisor. His latest book is called Lean AI: How Innovative Startups Use Artificial Intelligence to Grow, and it's currently available on Amazon for purchase now. Hello Lomit, how are you? Thanks for joining me on the show.
Hello Sean. It's great to be here.
Excellent. I'm really excited to talk about this topic. I know we've talked about it before, but obviously you have invested in it significantly. You have ton of experience in it as per the book. Before we get into that, I'd love to give you an opportunity to tell us a little bit more about yourself. Share that with the audience and talk to us about how you became the VP of growth at IMVU.
Sure. Yeah. Hi everyone. My name's Lomit Patel. I have pretty much worked in startups for over 20 years and my role has always been joining startups usually really early. I've always come in. I've been responsible for helping to acquire customers, figure out how to retain customers, and more importantly, how we're going to monetize customers. It's all around helping the company to ... Once they've got a product it's really, how are we going to continue to grow from there and really turn it into a meaningful business?
Been fortunate to have worked at several startups that have had successful exits, which are potentially defined as companies that have been acquired or gone IPO. It's always been great to learn going through those experiences. Currently, I'm working with a startup called IMVU, which is actually the world's largest avatar-based social networking app. It's like a mobile role playing game where you create an avatar and create a virtual reality world where you get to make and engage with different friends. I'm doing the same thing here where I'm head of growth, so I'm responsible for all of our acquisition, retention and monetization.
Thank you very much. I appreciate that. It's also helpful to know more background about IMVU. We're going to talk a little bit more about that and how you've used your expertise to grow efforts there. Shifting gears, we're going to just talk about the topic that we wanted to discuss today, and that's also the topic of your book, right? How to leverage artificial intelligence, automation for growth, more timely now than ever. If you could, give us some more background here about what it means and how you've used it to benefit IMVU.
Yeah. What I'll start off is just sharing my inspiration for writing the book. That primarily came when I joined IMVU just under four years ago. One of the things that I wanted to try and practice here was an area that I was really passionate about, which was all around AI and automation, which for the most part was being talked about, but not a lot of companies were doing that. When I joined here, one of the things that was abundantly clear is that we had so much customer data.
The truth is the customer data isn't valuable unless you're able to really extract value and take action on that to really help you pivot and make the right decisions to grow the business. Coming in here it was this turnaround situation because the company wasn't really growing in the right direction. A couple of things that I was able to do, but one of the big things that's had a profound impact on growing the business was to help transition IMVU into a mobile-first business.
Going into mobile, as you know and all of your listeners know, is that people don't have a lot of attention span in mobile. You're able to get a lot of data, but you have to really be able to take real-time action to really get users engaged with your app as quickly as possible. What I was able to do was to integrate all of our data sources because we had a desktop. We had a pretty successful desktop app, but it wasn't growing. Mobile was this new growth engine for us.
Really integrating the data source to really get a singular view on who our customers are. Once we got that, I was able to really figure out with that data, how can we get smarter and grow in the business? With that data, we're able to leverage AI onto that data to really draw extracts to answer questions like, who's the right users that drive the best lifetime value for us? How can we build more audiences like that across different channels where we're doing a lot of paid user acquisition? Like Google, Facebook and a bunch of other partners.
We were taking a lot of that data and then building API pipelines where we were able to extract that data back into these different partners and build better audiences. Then the question was, once you got the right audience, what's the right message? We're getting a lot better around personalization. People talk about personalization, but you know, we're able to really put that into practice because we're able to take data and use that data to really come up with the most relevant creative messages that would get users to take the actions to come into our product.
Then once they're in the product, the question was, how can you get users to engage deeper into the product to get them to really see the value proposition of the product? There, we were able to use the data to come up with personalized experiences based on where users were coming from and what actions and behaviors they were doing, so that in real-time, we were able to pivot the product to provide the most relevant experience to get the user to get deeper engaged in the product.
Then the other part of that was, once they were engaged the question was, how are we going to monetize these users? We had two different ways to monetize users early on. One was around doing either in-app purchases in our product or through advertising revenue. We were able to leverage AI to really help us to identify which of those two buckets a user was going to fall into pretty quickly. That based on those predictions, we were able to get them into the most relevant experiences so that we were able to figure out how to start monetizing those users as quickly as possible.
The high level is it's all about understanding the whole entire customer journey, but the heartbeat, it all comes down to data. It's about having the data to really help us to figure out what are right ways to engage with a user at different stages of their user journey that ultimately helps them to see the most relevant user experience, to get them engaged in a product and enables us to make sure that we are able to keep those users around so that we can monetize them.
I think you did a great job of laying out basically the framework and then the steps or milestones involved, which is super helpful because you never want to do the steps necessarily out of order, right? Then you use the tools. You leverage the tools like AI and automation to help you be more efficient and effective at doing so. You said a lot and I have a million questions for you.
With the time that we do have, where I'd like to start first is, can you talk a little bit more about how you use these technologies to, as you were describing, find those, how I would interpret it is, best-fit customers, and set up a better audience, is that the language you used, which I like a lot. It sounds a bit like product positioning, but I'd like you to correct me here or add a little bit more color to it because it's very interesting.
I know the differences between them, but I'd love to hear you articulate the benefit that you get out of conducting an exercise like that and how you go about doing it and using those technologies.
Yeah. One example of what we do is we ultimately aggregate all of our data to really come up with a good, better, best type of user audience for us based on lifetime value. An example of a good customer is somebody who comes and spends time in the product, but necessarily doesn't maybe spend a lot of money. The best customer is somebody who's highly engaged in the product and spends a lot of money so that we know that they're worth a lot more for us. Based on that we're basically building behaviors on what are the users that fall into those different buckets?
We build profiles on that data because once we build the characteristics of what goes onto those buckets, then we take all that user data and then we put it across the different platforms where we're trying to attract more users like that. Because a lot of these partners that we work with, whether it's Google, Facebook, different ad networks, all of them have user data, but what they need is they need to get the right data signals or insights from you so that they can use that to pre-populate and find really good matches of those types of uses for you.
For example, from our user data, they're able to extract the characteristics in terms of behavior. Are these people more likely to be students? Are these likely to be more male or female? What kind of age or characteristics? What types of interests do they have? A multitude of different data points come out of the data that we share with the different partners to build these different profiles of users.
Then the key thing is for us, because we have a scoring system, we know how much we're willing to spend to acquire users that fall into those different buckets, whether it's a good customer, a better or the best. They have different values associated with that. The other thing that they have associated with that is the messages are going to be different that are going to resonate with each of those different customers in those different buckets.
What we try to do is not only do we try to find those relevant users across the different platforms, but we also try to create the right types of personalized messages that are going to resonate as well. The truth is, ultimately it's all about running experiments. It's about trying to increase your velocity. You're trying to run experiments, because for example one thing that happens all the time is the business is really fluid. Things are going to change based on what's going on around internally and externally. Externally, for example, COVID-19.
That is a different situation where people have jobs or if they don't have jobs in terms of how they're going to respond to different ads. We have to continue to keep using the data, to keep coming up with different hypothesis on messages and look and feel and call of actions. All of that is just an ongoing virtuous cycle that's always on where we have our winning creatives that are always being tested against alternatives. Same thing goes with audiences.
We have a good, better, best, but then we try to break that down into more granular-level audiences within those buckets to see maybe an example being a good customer in the U.S. is going to be worth a certain amount, but a good customer in Brazil, which is another country where we ... because we spend money all around in the world. In Brazil, a good customer in Brazil is going to be worth a lot less than a good customer in the U.S. and is going to have slightly different characteristics based on the type of mobile devices that they use, the kind of bandwidth that they have.
There's a lot of changes that we keep doing, even culturally and based on languages and based on where people are coming from as well. That's the fun part about the job, which is that it's not just one and done. Every day, you pivot into whatever's going on, but the best way to do that is once you have the data and you have the AI providing insights, you try to automate as much of the tasks and processes. Tasks being the type of users you want to target, how much are you willing to ... You can make those adjustments on the beds of how much you're willing to pay for that customer.
On the creatives, you're running all these different experiments. The creatives are basically trying to ... The machine is trying to figure out, what's the optimal creative at any given time based on the right moment and the right part of the user journey that's going to resonate the best? Long story short, with automation, and maybe this number might seem impressive or not, but when we were trying to do this stuff manually when I first started, we were probably running a couple of hundred experiments. That was considered pretty good considering our team was probably like eight to 10 people at the time.
Now, with machines, I have a team of probably half that size, like four to five people, but we're running like five to 10,000 different experiments a month because the machine does all of the heavy lifting now and the people on my team. The people are really there to really support the machine, to put the inputs in and the machine is doing all of the execution, which used to take a lot of human manpower to do before.
That is a significant increase. Like you said, even cutting back on the resources that you needed to do it and an excellent way to leverage the technology. As in, I like what you mentioned before, and I'm going to call more attention to that, it's all about running experiments, right? Increasing your velocity there, what you can do with the data that you have. Then conducting experiments with that data based on what it's telling you. It's very interesting because it just creates all these new opportunities to learn and other things to do with that data.
Like you said, everything from setting up the infrastructure to give the ability to do it, to your team members adopting the culture that this is the process, right? I've often seen too many instances where companies will make a decent amount of progress, but they will treat these lessons learned almost as milestones and as a static component, but it really is dynamic. You've articulated that well too when you call attention to external market factors like the pandemic and having to respond to that.
I remember when we first entered that phase, the processes a lot of companies were following were not effective any longer because things had changed and they had changed so quickly. I have to imagine that the technology will also help you respond to those relatively quickly. Can you talk about that as well, too? I'd love to hear more from you in terms of how the tech helps you prepare for what's coming and then how to think about using it in the best way to make sure that you're getting the value out of it.
You're not just introducing tech to introduce tech, because you can over-engineer anything, but how do you keep that core focus on lessons learned and making sure that it's moving the needle?
Yeah. What I would say is, with any tech, one of the things that you really want to use as a good starting key, as a starting point, is really clearly articulate what are the use cases that you're trying to solve with that tech? For us, one of the use cases that I wanted to pick early on, because ultimately trying to bring in digital transformation in any business is going to have resistance, because as humans we always are comfortable with what we know versus what we don't know, right?
One of the best ways to try and get universal support because ultimately, you need universal support to try and bring in any form of transformation, is to try and do it where you can get payback pretty quickly, where it can show value to the business. Working in growth is great because most companies always want to grow. Nobody ever wants to not grow, because if you're a company that's backed by the venture capitalists that's the mantra. It's pretty much grow as quickly as possible.
One of the ways for IMVU, and IMVU is probably similar to a lot of other tech companies, is one of the channels where you can really drive a lot of meaningful growth and you have a lot of control over is paid user acquisition, which is where you actually go and spend money on Google and Facebook and all these different digital channels. Because ultimately, you know that for each dollar you're spending, there's a certain return that you're going to get for that money and how long is it going to take you to recoup that money back.
The best way that I look at what we do in that use case of paid user acquisition is that we're day traders. That's what I keep telling everyone in my company, is ultimately, you know what? Because we have the second largest, if not the largest budget outside of payroll that we get to manage. As I say, there's no such thing as free money. If we're managing money then there are certain goals we have to hit.
For us, the two goals, which is really similar to most companies when it comes to growth, it's around acquiring new paying customers, I guess, around what's your ROI or your return on ad spend? Then the third part to that is, how long does it take you to recoup that money back that you're spending? With that use case as the backdrop, what we ended up doing at IMVU was ... And I was fortunate because I had a lot of good relationships with a lot of these companies like Google, Facebook.
Even now, people are talking about companies like TikTok and Snapchat being major advertisers. Nobody talked about them four years ago, but I had relationships there, so what I was able to do was coming in early at IMVU, was we became beta testers for a lot of these partners outside of Google and Facebook because we'd get into a lot of their alpha and betas. I'm talking about Snapchat. When Snapchat was trying to figure out how to monetize all those users before they went to IPO, we were one of the early beta advertisers.
We were able to help them figure out how to build an advertising machine. All of these partners, whether you realize it or not, have AI built into their platforms. Because they have so much data, they want you to spend more money with them. They want to get better and smarter around how to help you acquire your customers. The only difference is that they only look at things in silos. It's like, how are they doing against themselves? Because they have no preview into how they are doing against each other.
That was a use case that I wanted to solve because we knew how everyone compared because we had the universal preview into how all those channels compare to each other because we had all that data. Clearly, you're not going to share that data with all your partners because that's your secret sauce. Having said that, having been early betas in all these different partners, we also knew how their tech machine worked in terms of how their thinking was happening.
That's how we were able to automate a lot of levers around bids, budgets, creatives, because we know those are the things that we could control on their AI. We ended up building an AI machine that pretty much was ultimately about helping us to manage our budget as efficiently as possible to hit our goals, and to automate the levers that we could across all these different partners in the ecosystem that we already had good knowledge on how their training models worked with algorithms.
Because ultimately it's all about all the algorithms. It's about building an algorithm that can enable you to be a smarter algorithm to work these other algorithms of better machines. That's what we ended up doing.
Yeah. That's super exciting and understanding how ... so leveraging the tools in the right way to make sure like you mentioned, focusing on those use cases is important to ensure that you're driving towards that. You talked about thinking about it in terms of experiments and payback period. I like that as well too, because now you know you're not losing a little bit on every sale and making it up in volume, right? That doesn't scale.
Instead, those experiments have to pay off otherwise you're going to need to adjust to the experiments or run different experiments. The tech sounds like an excellent way to be able to do that. Thank you for sharing that background. That's a lot of context.
Yeah. Then one other thing I was going to add to that is, people probably don't realize, but all of these different ad partners that you work with, whether it's Google, Facebook, and the reason why I come back to that analogy of saying that we're day traders is because it's like the stock market. It's all based on supply and demand. The amount you end up paying in terms of CPMs or cost per clicks, all of that is really determined by how many advertisers are competing for that given impression at any given time? It's like the stock market.
If a stock is really popular, the price is going to go up. If it's not, the price is going to go down. This is the thing that people didn't realize, but what we ended up building was an AI machine that was looking at all of those data signals on supply and demand across these different exchanges. Because we knew how much we were willing to pay for a customer that have fell into that good, better, best customer segment that I talked about.
Instead of just giving all our money to Google and Facebook, which is what we used to do and how most people are doing it right now, and then you let Google and Facebook figure out how to spend your money. We were looking at like 20 to 40 partners at any given moment in time and looking at, where are we going to be able to get that good, better, best customer? We were pivoting our budget basically like day trading.
Moving it from one partner to another partner based on how much competition was going on in those different exchanges to find the optimal place to spend that money to find those customers. That's the thing that really profoundly changed the whole trajectory of our, business because we ended up going back to our goals. The goal is to like cost to acquire customer went down over 4X, which was amazing. It's continued to go down while our ROI went up over 4.5X over the last two and a half years.
The most important piece is the payback period, because ultimately you only get money to spend if you can recoup it back after a certain period. The average for most gaming apps is around like six to 12 months. That's where we were, close to six, when I started four years ago. Now, we're at less than 30 days. Within 30 days we recoup back a hundred percent or more of our user acquisition budget. We're pretty much in the business of recycling money and buying growth instead of going out to raise money, which is a big challenge when you're a startup.
All we're doing is we're basically just using the same amount of money that we had and we just end up buying more growth with that because we don't end up losing that money because we recoup that money back. We have a high predictability now because our algorithms have continued to get smarter over time.
Yeah. That's some incredible numbers. The math really makes a major differentiator there when you're talking about being able to fund your own growth, like you said. I imagine the technology is realistically making that much more feasible and possible. I'm always an advocate for whenever you can fund your own growth, do so if possible, but obviously you need the numbers to be within a certain range in order to be able to do that.
Yeah. It's like anything. Yeah. As you continue to get smarter around ... Going back to what you were talking about, it's all about understanding your customer and understanding the evolution of that customer, because whatever customer you started off with in your business, let's say four years ago, is going to continue to evolve because people change and the product can change. It's all a matter of trying to find the right product market fit at any given time as the product and the customer continue to evolve.
For sure. Well put. A question I have for you then is ... And I'm sure some of the listeners are curious about this as well, too. Given the nature of the amazing growth and progress you've made using these strategies at IMVU, for those that are running less experiments with larger numbers of teams involved or it's heavily manual, right? They want to get started in figuring out what they can do to leverage these technologies like automation and artificial intelligence in order to grow and start to conduct these experiments, what would you recommend they do? How would you recommend they get started?
What I would recommend is going back to what I said. All of these different partners where you're probably spending money right now, already have some form of AI and automation built in. The only difference is it's built in, in silos to be self-serving for them, because ultimately it's about you spending more money with them. The starting place is really make sure you've got your data aggregated into one place because data is the heartbeat. Without data on your customers you can't really do anything else when it comes to AI and automation. Make sure you get your data.
There's words like CDP, which is like customer data platform, but the idea is, integrate your data into one place, ideally in the cloud, so then from there you can start putting it into these different partners. To get started, I would probably start with maybe like Facebook and Google if you're a small business, to at least give them the right data and see what they can do for you.
From there, I mean, just to give you rough numbers, you want to be spending probably like 75 to a hundred thousand dollars a month to make it meaningful before you start thinking about maybe trying to build or buy some form of AI intelligent machine for yourself. What I would do having gone through this whole process now, is that, when we did this, there wasn't really anything that really existed out there at the time.
I can tell you a little bit more about my journey, but ultimately we ended up buying it, but we did it in a way where we partnered with another startup that ultimately was building something like this. But as you know, startups are always trying to build something, but their biggest challenge is to try and figure out if it works. I ended up working with a startup that was building something of what I wanted. I ultimately had a vision.
I created a pretty detailed product spec of what I wanted this thing to look like and partnering with a young startup, I was able to influence and get them to really pivot their business and their resource to build that spec of what I was looking at and which is what I call today is the intelligent machine. With that, and with all of my relationships, we were able to get all of the APIs from all these different partners, because we were able to get into betas with a lot of these partners.
We were able to build the API connectors into this machine, where it was able to then automate a lot of those levers of changing bids, budgets, creatives in real time. Then we didn't really need to have a lot of people on the team running and doing all those manual tasks and processes. I was able to make the team a lot leaner in the process and focus more about bringing in people that were going to just support the inputs on making sure the machine was going to work as well as possible.
Long story short, start off with the different partners right now. Give them the right data, see what they can do so you have a baseline of what you can get. Then from there, try to figure out if you want to go ... I wouldn't recommend building anything at this stage because this technology is changing all the time to what we started to where we are. You want to ideally work with a SaaS platform now that is more dedicated into building this because it's ultimately, then they've got the resources to focus on trying to build that product.
You just end up becoming a partner that can leverage that technology to see if you can customize it to a certain extent, to help build this new layer of AI that can sit between you and these different partners that you're working with. I'm happy to give you the name of the company that we ended up working with because they have actually become a platform that they're now starting to offer to our partners. You can piggyback off a lot of the stuff that we ended up doing there.
I think that's a perfect place to transition into talking about resources. Nicely done. Thank you for sharing the framework. I think that's super helpful for anyone who wants to get started. I want to thank you obviously for being here and sharing your incredible knowledge with myself and our audience. Before I let you go, two more questions before we wrap. The first one as I just mentioned is, what resources, including the one you just mentioned, would you be interested in sharing with our audience?
Yes. In terms of the resource, I would highly recommend people that are interested in trying to figure out how to build an AI intelligent machine, but more importantly figure out a SaaS partner that can provide this for them. Nectar9 is the name of the company. N-E-C-T-A-R-9. That's the partner that we worked with at IMVU. I would say that that AI-intelligent machine that they have now can pretty much solve almost like 99% of all the different use cases any marketer would ever need when it comes to growth marketing. Because we use it pretty much for an entire user journey, from acquisition, retention and monetization.
It's gone through a lot of the learnings that we ended up going through. It would enable any company to fast forward and be able to piggyback off and without having to go through a lot of the pains that we went through early on. The other thing that I would recommend in terms of resources, I write a lot about this topic on LinkedIn. I would encourage people to follow me on LinkedIn. It's pretty easy. I connect with anyone who reaches out to me. Reach out to me, Lomit Patel on LinkedIn.
I also have a blog where I publish a lot of content too. My blog is really simple. It's my name, Lomitpatel.com. Between my blog, LinkedIn, it's really easy to get ahold of me. If you want to message me through LinkedIn or there's a way to reach out to me from my blog as well. Then, last but not least, the book that I have, Lean AI, it really got into a lot more detail in terms of the entire journey on how to really bring that transformation around AI and automation into your business.
The book's kind of broken up into two parts. One part is around the AI and automation. The second part is all around growth marketing strategy. Once you have the AI the other piece to that is, what are the different use cases that you ... And based on different businesses have different strategies. It really talks about a whole framework of how to figure out the right strategy to help you grow your business. That's really based on over 20 years of my experience, having gone through and worked through a number of different types of startups.
Because ultimately, there's a word where people come up with called playbook, but the truth is, there's no universal playbook that works for every business. You have to create a unique playbook, but the components of the playbook can be applied. It's like a jigsaw puzzle. What the book talks about is what's the right components that you need based on the challenges at your businesses and the stage of the growth of where you are. Because that playbook will continue to evolve as your business evolves and grows.
Fantastic. Thank you for breaking that down for us. As I mentioned earlier on the show, the book's available on Amazon. I will link to that in the show notes, as well as all these other incredible resources you provided. Thank you a ton for that. Thank you so much for being here and sharing your incredible knowledge with myself and our audience.
It was a pleasure being here. Thanks for having me Sean.