ABOUT: LinkedIn  | Twitter | Lean AI Book | IMVU



ABOUT ROCKETSHIP HQ: Website | LinkedIn  | Twitter | YouTube

KEY HIGHLIGHTS

💸 Why spend money on retargeting when you’ve already spent money on acquiring users.

3️⃣ The three key segments Lomit recommends retargeting – and why.

📧 Why push notifications and emails can’t reach all the users that you might want to retarget.

👬🏼 How Lomit recommends setting target KPIs for retargeting different segments of users based on their behavioral profiles.

📈 What incrementality is – and why it matters.

🤔 Why Lomit recommends going after different sets of retargetable users on different channels.

📱 Why IMVU shifts communication from paid retargeting to emails & PNs after a user’s first 7 days post-install.

👨‍💻 The user behavior inputs that let IMVU use AI to build personalized creatives.

🚋 How deeply understanding the ideal user journey helps IMVU craft personalized ads for users at each stage of the user journey.

💯 How Lomit thinks about building vs. buying AI technology – and how maintaining an in-house system can be a challenge.

KEY QUOTES

Identifying highest priority users

If somebody has installed the app but hasn’t registered, that is a high priority because, until they register, we can’t really do any of the other things that we want. A big part of the focus is really the first seven days. 

Other factors supersede ROAS sometimes

ROAS is definitely the key variable that we try to track, but based on which segments that we use in retargeting ads. For example, when we’re targeting people that have installed the app, but haven’t registered, there’s no point really looking at ROAS at that stage, there we’re kind of looking at the conversion rate to getting them to become a registered user. 

How to come up with incrementality

When we look at the ones that we’re targeting on the DSP or on Facebook, we always try to run a control holdout group. And then, so generally what we do is when we look at the holdout which are people that don’t get exposed to any of these ads, how do they end up behaving or how do they end up monetizing versus the ones where we are paying, and that’s how we sort of look at the delta to come up with our incrementality.

FULL TRANSCRIPT BELOW:

Shamanth: I’m very excited to welcome Lomit Patel to the Mobile User Acquisitions show. Lomit, welcome to the show. 

Lomit: Thanks Shamanth for having me. 

Shamanth: Absolutely. Lomit, I’m also excited to have you here because you’ve already been on How Things Grow, and much of the story you told on How Things Grow was about really the evolution and the start you and your team got on mobile, after having been a web only company; and it feels like what we’re going to talk about today is going to be a logical progression of that in terms of how far you guys have come since you guys embraced mobile for the first time. It’s so like you to really be at the leading edge today as you were back then. So I’m very excited to dive into a lot of what we’re going to get into today. 

Lomit: Definitely, let’s do it. 

Shamanth: Yeah. So, Lomit, to, you know, we are going to talk about how reengagement and retargeting ads can be effective across the user funnel. How do you decide what user segments to retarget via ads because, if somebody came in, never made a purchase versus somebody who made a purchase and didn’t come back for 90 days – these are very, very different kinds of people – how do you decide which one to go after, which one could be better, because presumably there are an infinite number of segments you could go after? 

Lomit: Yeah. So the way we think about it, and, a lot of people now are talking about retention, reengagement, and retargeting sort of being an important part of the entire growth strategy, we’ve been doing it for a while right from the beginning because the way we looked at it is, just focusing on acquisition without retention doesn’t make a lot of sense, because we’re spending so much money on the acquisition side, it’s kind of a sunk cost, because that money is kind of gone at that stage. 

Industry standards, good conversion rates, so generally, whatever registration rates like 10 to 15% or, you know, conversion rates to like payers being like 1 to 3%, it’s very low when you think about it, you’re kind of leaving the majority of people on the table that you haven’t been able to convert. 

And so our philosophy has always been, how can we squeeze more out of every, every user that we’re bringing in, because ultimately we spend that money to bring them in, and if they don’t, if they don’t convert, then it’s the loss to us as a business. And so our philosophy, when it comes to looking at retargeting is important at a couple of stages which we identify. 

One is, you know, when somebody sees your ad and clicks on it and comes and installs the app, but then they don’t end up registering, that’s one second that we try to go after because we know they raise their hands, they showed some interest. Clearly, they didn’t follow through, and so we want to try to get them to at least follow through and register. And once they’ve registered, the second part that we look at, just like a lot of other mobile apps, is do you really have a small finite window that we call the onboarding process, which for us is the first seven days. 

But, we know, if we don’t get users engaged into the app within the first seven days and really truly experience the value proposition as quickly as possible, then they’re going to be more likely to abandon and never come back. So that’s the second segment we focus on it is the first seven days. And then beyond the first seven days, the other thing that we continue to track when I use is how engaged they are or if they have lapsed and, and have become dormant. And so the third segment we focus on is lapsed users where we try to do win back campaigns to continue to try and bring them back into the product and use it. 

Shamanth: Got it. And how do you identify that these are the highest priority users to go after? 

Lomit: Right, so for us, clearly talking about it in order,

If somebody has installed the app but hasn’t registered, that is a high priority because, until they register, we can’t really do any of the other things that we want. A big part of the focus is really the first seven days. 

One of the challenge with CRM campaigns that we find and I’m sure this probably applies to a lot of other advertisers out there is, you know, CRM campaigns are highly dependent on how many people opt into your emails. 

Shamanth: Right. 

Lomit: And, for us, when it comes to email opt-ins, we have double opt-ins because you want people to engage with our email, but also opt into permission so we can continue to message to them. 

Shamanth: Sure. 

Lomit: And so, opt-in rates are okay but they’re not huge, because we have a few frictions and barriers. The other thing that we do is, we try to get people to opt into our push notifications and, as you know, industry standards on push conveying between like 20 to 40 to 50%. For us, we are more on the lower end of that spectrum, because, a lot of people again, what we find is people opt into push after they’ve actually come into our app, and, and they’ve actually started doing some behaviors and actions which are more contextually relevant. 

So, people are more likely to opt into push in our app once they have started making friends and they want to get notified when their friends are coming back online, or if they’ve shown interest in different chatrooms and when there’s activity happening in those different chatrooms or if they are showing interest in purchasing certain items, and, when there might be a special deal on that. 

So those are contractually relevant, and once we hit people up to opt in at those stages, we get a much higher opt-in than try to during the onboarding flow. And, so for us, retargeting ads are like, the best way for us to message users early on, because with retargeting ads whether they have opted in or not, we could reach the large majority of those users outside of the app. 

Shamanth: Understood, understood. And that makes complete sense just because everybody hasn’t experienced the app in its entirety to really get value out of the push or emails or even be accessible by the push or emails. So retargeting really lets you cast as wide as possible a net, and actually get these people to start experiencing the app as much as possible. Right?

Lomit: Right, definitely. 

Shamanth: Yeah. And, how do you set target KPIs which could be a ROAS, I imagine, for different segments of users, because again somebody who hasn’t registered probably would have a different ideal outcome than a dormant purchaser – how do you think about that? 

Lomit: Yes, as you mentioned,

ROAS is definitely the key variable that we try to track, but based on which segments that we use in retargeting ads. For example, when we’re targeting people that have installed the app, but haven’t registered, there’s no point really looking at ROAS at that stage, there we’re kind of looking at the conversion rate to getting them to become a registered user. 

Right, and then once they’ve become a registered user within the first seven days, you know, that’s when we start looking at two KPIs. One is the conversion rate to becoming a payer, because that’s, because within the first seven days we, at least for us, if we can get somebody to make a nominal purchase in our app, because generally what we do is we sell credits which is IMVU currencies that people buy to redeem against different purchases. 

And so, if we can get somebody to make a nominal purchase like $1 worth of credits then, that’s the other conversion rate we look at within the first seven days. And if they don’t do that, then the other thing we try to look at is the other way we monetize users is for advertising. 

And so, generally through advertising, we get people to take different like surveys or watch rewarded videos to earn these credits. So, we’re able to identify how we’re going to ultimately monetize these users, and based on that we try to we track the ROAS based on the different segments that they might potentially fall into.

Shamanth: Understood. So it’s almost like you look at the monetization profiles, and the target KPI for them is a function of the monetization profile they’re expected to have. 

Lomit: Right. And then with lapsed users that have, gone dormant, with them, we generally try to look at we continue to track ROAS, but it’s like, you know, can we get them to come back into the app so we kind of look at engagement rates there. 

Shamanth: Understood. 

Lomit: And time spent on the app, yeah.

Shamanth: Understood. So do you look at the ROAS of your reengagement campaigns in isolation to say: right, I spent so much on dormant users, let’s make sure we make so much back or is that also tied with your acquisition budgeting as well? 

Lomit: So generally, it’s all tied, in terms of – how much are we spending and then all of these activities but, but we do, to what you were saying, we do look at each of those activities in isolation as well.

Shamanth: Understood.

Lomit: Just to make sure that there’s incrementality coming from those different spends that we’re doing. 

Shamanth: Understood. And that incrementality, it’s certainly a phrase that’s used quite a bit. Tell us what that means, tell us why that’s important. 

Lomit: What it means to us is, is that, for every dollar that we’re spending, is it actually leading us to find incrementally new users or payers that we wouldn’t have got ordinarily without spending that money. And so generally what we try to do is, whether it’s spending, we are able to identify what the baseline is based on what that spend is on the different channels or the different activities across like, in this example, it’s retargeting, because we’re retargeting to use a couple of different partners, one would be Facebook, and then, another partner could be a potential DSP programmatic partner that we’re using. But what we try to do is we try to split our audiences up for retargeting, so the ones that go to Facebook, we try to isolate those users not to go to the DSP, so that the people that the DSPs are targeting are not going to be the ones that we are ordinarily going to target on Facebook anyway. 

Shamanth: Right.

Lomit: And, then

When we look at the ones that we’re targeting on the DSP or on Facebook, we always try to run a control holdout group. And then, so generally what we do is when we look at the holdout which are people that don’t get exposed to any of these ads, how do they end up behaving or how do they end up monetizing versus the ones where we are paying, and that’s how we sort of look at the delta to come up with our incrementality.

Shamanth: Understood. So it’s pretty much a controlled AB test that you’re doing. 

Lomit: Right. 

Shamanth: And the way you describe it, it does take, I imagine, a fair amount of resources. So can you tell us what sort of apps is retargeting not right for, if any?

Lomit: So, just going back to what I’d said, if you’re doing paid user acquisition and you are not doing retargeting, it’s already a sunk cost at that stage, because you already spent that money. And so, if you’ve already spent that money and you, generally the more you’re going to keep putting into acquisition, you know, and your conversion rates, it’s still going to be in under 10% at the most unless you’re like, doing something that’s hugely different from the industry standard. So you are still in the vast majority.

 So my philosophy is that you should be considering retargeting at least within the first seven days because a lot of people aren’t going to opt into a lot of the other communication channels just to try and get them engaged into the app. And what we try to do is once, after the seven days, we tend to spend a lot less on our retargeting paid efforts, and, and increase a lot of our in-house communication channels, because by that stage we get a lot more people, by then we have a pretty decent number that opted into our push, pretty decent number they’ve opted into our email, plus, you know, people are starting to visit the app frequently, so you could use in-app messages too. 

Shamanth: Understood. What I also find fascinating about how you and your team have approached your entire reengagement effort is you guys have been incredibly sophisticated in identifying users at different stages with different personas and you have used AI based testing for doing creative testing and improving on creative performance. Can you speak to your process of how you have done this AI based testing? 

Lomit: Sure. So where AI has really helped us, because for the most part, there is so much work, as you mentioned, that really goes into doing a lot of these granular testing and segments holdouts and, and orchestrating all of these types of campaigns, we try to automate a lot of that process. So, there’s not a lot of manual intervention that’s involved in orchestrating and executing these campaigns, but a lot of that is driven through data insights, and that’s the part that we’ve really leveraged a lot of AI into. 

So all of our data where we are spending the money and then how the user is behaving lives in this cloud storage, and then we try to apply AI into that data to try and surface insights to try and help us to identify what are the behavior actions that users need to take within different stages of these retargeting and re-engagement campaigns that are going to be leading them to take the best next step which we describe as the ideal user journey.

And so, once we start seeing certain patterns that start bubbling up, that’s, that’s, you know, where we sort of take those insights and then try to come up with personalized creatives to try and match those behaviors. And so, for example, for us, we’ve kind of mapped out these ideal steps. The first is like somebody has to create an avatar, you know, once they come into IMVU because if you don’t have an avatar, then that’s really your identity, then you’re not really going to get to participate in our social network. 

And then once, once you’ve created an avatar, we always give people like free credits, and we want people to start redeeming those free credits, because the free credits kind of takes a basic avatar and helps you to customize it to your look and feel. For example, the look, your eyes, your skin tone, the clothes you want to wear, and then after that, we know at the end of day, our key value proposition is really about connecting people and making friendships. So, once they’ve created an avatar, they’ve started redeeming their credits to dress it up, we want to start getting people to start engaging in our chatrooms, start making friends, and start exploring a different shop to try and see what else there is out there, and start engaging in our feed so that other users can get to know them as well. But it’s kind of creating that social viral loop at the end of the day.

Shamanth: Right. 

Lomit: And, all of our creatives lead people into that loop and continue to get them to rinse and repeat those behaviors. 

Shamanth: Understood, that’s fascinating. So it’s almost like, you know for instance, at the end of day three, they should have added a friend, but they haven’t, and they should have bought certain avatars or they should’ve bought selling articles of clothing, and they haven’t, therefore – and that’s what your AI algorithms are telling you because like, look, this is an ideal user journey, users don’t have the ideal user journey, let’s show them creative that lets them take that next step. 

Lomit: Right. Perfect. And so, another way to kind of look at our AI is basically the sherpa, it’s basically the guide. And, it’s the guide but, the way it does communications is through the creatives.

Shamanth: Yes, yes, that makes complete sense. And I understand you guys use a third-party AI technology, how do you recommend that marketers think about building a system in-house versus potentially getting a third party partner? 

Lomit: I know companies always talk about, build versus buy, and for us ultimately, the decision to end up going with a partner really came down to the fact that, it takes a certain type of skill set to really build this, and to have that level of expertise in, in data science to – because it’s not like, one and done kind of thing, because, because all of this is just, algorithms that continue getting trained by the data, but you still need, sophisticated data scientists to continue to validate the model, make sure there’s no biases happening in there that you continue to attract. 

And so we don’t have that level of expertise in-house which is why we ended up going with a partner, plus my personal experience and I’m sure a lot of other people might be able to relate to this, I’ve worked at companies where you build a lot of these things, especially on the CRM side, because, the commonsense is, we can build a lot of these things which is great but what people tend to forget is once you build it you have to maintain it and you have to continue to iterate to keep it going with best practices, and that’s where a lot of companies don’t end up putting the resources into maintaining it and iterating it, so it continues to stay on best practices. And what started off sort of becoming a competitive advantage, ends up becoming a liability in the long term.

Shamanth: Indeed, indeed, and that’s just because it’s not your company’s core capability.

Lomit: That’s right.

Shamanth: You might as well work with somebody whose core capability it is. 

Lomit: Right.

Shamanth: Indeed Lomit. Lomit, this has been fascinating and, as always, every time I talk to you, I’m so impressed by everything you’re up to. It’s amazing to dive into everything you’re doing with AI including what you have done lately to really teach the rest of us about how the work you’re doing in AI. Would you like to tell us about it before we wrap? 

Lomit: Sure. Yeah, so Shamanth, always a pleasure talking to you as well. And, everything that I talked about on the show today, plus a lot more, I’m really passionate about the whole area of AI and automation and, how to really leverage that, especially when it comes to startups where you have like, less resources but trying to get more out of that. 

So I ended up writing a book called Lean AI which is part of the Eric Ries’ The Lean Startup series, and it really goes into a lot more details to really provide the roadmap for any companies that are really looking to implement AI and automation to become their competitive advantage, to scale and grow their business. And it’s available on Amazon, so, love for your viewers to check it out. 

Shamanth: Indeed, indeed. We’ll definitely link to the book in the show notes, and I’m very excited to check out the book Lomit. Thank you so much for appearing on the Mobile User Acquisition show. 

Lomit: Thanks for having me Shamanth, always a pleasure. 

A REQUEST BEFORE YOU GO

I have a very important favor to ask, which as those of you who know me know I don’t do often. If you get any pleasure or inspiration from this episode, could you PLEASE leave a review on your favorite podcasting platform – be it iTunes, Overcast, Spotify or wherever you get your podcast fix. This podcast is very much a labor of love – and each episode takes many many hours to put together. When you write a review, it will not only be a great deal of encouragement to us, but it will also support getting the word out about the Mobile User Acquisition Show.

Constructive criticism and suggestions for improvement are welcome, whether on podcasting platforms – or by email to shamanth at rocketshiphq.com. We read all reviews & I want to make this podcast better.

Thank you – and I look forward to seeing you with the next episode!