Subscribe to the Growth Colony Podcast on Apple Podcasts | Spotify | Google Podcasts | Amazon Music | Stitcher | Acast | Pocket Casts | Breaker | TuneIn
Episode’s topic: How to calculate your LTV?
In this episode, host Shahin Hoda chats with Oren Cohen, COO at Voyantis, about how marketers can calculate their customers’ lifetime value (LTV) and leverage it to make better decisions.
Oren starts with defining LTV and gradually dives into the components marketers should bake into the traditional formula to arrive at a meaningful number. He then shares the importance of predictive LTV calculations, emphasising that the proper model selection based on the business problem is critical for better insights.
Oren concludes the discussion by talking about the LTV-to-CAC ratio and warns about some common obstacles that hamper LTV-based decision-making.
This episode’s guest:
Oren Cohen, COO at Voyantis
Oren is the COO of Voyantis, a no-code platform that helps PLG companies boost their paid marketing ROI by predicting the LTV of their users. Voyantis uses ML to activate zero and first-party data to target the right customers from the beginning. Early adopters of Voyantis are some serious players such as Miro and Notion.
Before joining Voyantis, Oren was Iron Source's VP of Demand and Operation.
On a personal level, Oren describes himself as an optimistic vegan who is excited about solving growth challenges using AI.
Connect with him on LinkedIn
Conversation segments on this episode:
- [02:09] Defining the lifetime value (LTV) of the customer
- [04:16] LTV starts from zero & plateaus at a later stage
- [04:50] Components of calculating the LTV
- [05:01] Account for all sources of revenue, base population & the accuracy of the evaluation method
- [14:40] Predictive LTV and its applications
- [16:30] What is a good CAC?
- [17:13] Cashflow is your limitation and the payback period is your goal
- [17:20] CAC is not your goal. You should look at CAC as the outcome
- [18:13] CAC for sign-up will be different than CAC for a subscription
- [19:15] Look at LTV to CAC ratio. For B2B companies, look at the CAC for the “aha” moment
- [22:13] What's a good LTV to CAC ratio? In the D2C world, it is 3:1
- [23:45] How to capture data for an accurate LTV calculation
- [26:08] Defining first party and zero party data
- [27:11] Obstacles to building good LTV models
- [31:49] Where should marketers start with LTV predictions
- [38:21] Advice for B2B marketers - adopt LTV predictive modelling techniques
Resources mentioned in this episode:
- About Voyantis
- About xGrowth
- Man’s Search for Meaning by Viktor Frankl - Book recommended by Oren
- Mobile Dev Memo - Resource recommended by Oren
About the Growth Colony Podcast
On this podcast, you'll be hearing from B2B founders, CMOs, marketing & sales leaders about their successes, failures, what is working for them today in the B2B marketing world and everything in between.
Growth Colony is produced by Alexander Hipwell and Allysa Maywald from xGrowth
It was edited by Dave Somido with additional editing by Allysa Maywald and music arrangement by Alexander and Allysa.
Special thanks Teena Wabe, we couldn’t make the show without you.
Growth Colony is hosted by Shahin Hoda, Director of Growth at xGrowth.
If you enjoy the podcast, please subscribe, and give us a rating on Spotify or Apple Podcasts.
Get in touch!
We would love to get your questions, ideas and feedback about Growth Colony, email podcast@xgrowth.com.au
Episode Full Transcript
[01:15] Shahin Hoda Hello, everyone, welcome to another episode. I'm Shahin Hoda with xGrowth. And today, I'm talking to Oren Cohen. COO at Voyantis, about how should marketers in SAS companies calculate the lifetime value of their customers. And what are some of the pitfalls they should avoid. On that note, let's dive in. Oren. Thanks for joining us.
[01:37] Oren Cohen Hey, Shahin, thank you so much for having me. Really excited.
[01:40] Shahin Hoda Absolute pleasure. I am I was just saying before we start recording, I'm super excited for this podcast. I think this is going to be, this is an interesting topic. This topic has been around for a while. But I also think a lot of people, a lot of founders don't completely grasp, but there is there's some technicality to lifetime value calculation and LTV for short. So I'm super excited about this. And I want to start with defining LTV, lifetime value, or what is your definition of LTV? How do you define that metrics?
[02:16] Oren Cohen Yeah, great question. So I think at the end, if we were looking at, let's say the somehow dry definition. So LTV will be the estimate of revenue that the customer will generate throughout of their lifespan as a customer. And these words of the customer can help determine many economics decision for a company, right? So including marketing, budget, resources, allocation, profitability, and forecasting. But, I must say that it's never about just the LTV. It's about the LTV at a given point of time, right?
[02:47] Oren Cohen So it's about the LTV curve and how it's developed. It's about the how companies cumulate the LTV, and it's about slicing by the relevant dimension for you. So it can be the LTV of subscribers or signups. But again, it's really about the LTV per specific goal for which you are aiming. But this is just from a very dry definition. Probably there are more complicated models for that, but we'll discuss it in a few secs, probably.
[03:15] Shahin Hoda I love it. And I mean, you talked about the LTV curve, and traditionally LTV tends to be, so I mean, that's a very interesting point. I mean, there's a time component in terms of the lifetime of the startup component and then just like you said, there are taking that into consideration from different angles of what that is measure and the LTV curve is usually it starts high. And the objective is to reduce that over time. Is that Is that correct?
[03:41] Oren Cohen No. So I will say that LTV starts with zero, right? Because basically, at the beginning, you have no idea what will be the LTV of each of your customers or clients. But the question is, first of all, what's the time period that it will take you to recoup or to be at the break-even point at the end startup is big companies are being limited, first and foremost by their cash flow. So you're looking at your cash flow, and now the question is which of the payback time is ideal for you?
[04:12] Oren Cohen Oh, what's the maximum payback time you can allow yourself? So I will say that LTV starts from zero and grow upwards. And then obviously there is in most of the product or services, some kind of a plateau. So this is probably the behavior but I think looking at the curve, or what will be the payback overtime, lets you as a start up a company deal with the cash flow limitation to maximize the value that you're being able to generate and obviously to define what's your margin target. right?
[04:44] Shahin Hoda Got it. I totally see your point and where I went wrong there. Tell me a little bit about the components that go into LTV calculation.
[04:55] Oren Cohen Yeah, so this is a very interesting question. So first of all, let's again start from the basics. So, probably the basics will be the total revenue stream from all sources. So we are talking about subscription, virality effects, referrals, upsell, cross-sell everything, which has some kind of an influence on the income that you see as a company. The second component I will look at is obviously, what will be the base population. And this is what I mentioned earlier. So, denominator is what defines the group of which you measure and act. So it can be the LTV of your signups, subscribers, the free trial. So it's really, if the first part was dealing with the revenue, the second part is dealing with the base population.
[05:40] Oren Cohen Now, the third thing, which is extremely important, and many companies just forget this, is the evaluation method. if you know what's your streams of revenue, and you know, what's the base population, yeah? Now you need to make sure that you have in place, defined a well-defined method to evaluate the model that you use in order to predict the LTV. One thing which is extremely important is when you're dealing with not the actual LTV, the accumulated revenue so far, but you're dealing with predicted LTV. So it's also what's the actual timeframe.
[06:17] Oren Cohen So it's really about the use of the LTV model. But let's say that we want to use this LTV model to improve our user acquisition, or ad spent investment in general. So we need to also look at the limitation on the network side. For example, seven days attribution for Facebook. So I want to make sure that my prediction or the insights that I'm getting regarding my LTV will be accessible during this timeframe that I can still act. So what is the max timeframe, you can update your prediction and still act upon it.
[06:51] Oren Cohen Now, there are obviously some advanced nuance when it comes to components. So we're looking at monthly commitment or collectible revenue vests commitment, chargebacks, refund, quick cancellation. So all of these if we are looking at LTV, we also can have a higher resolution and look at the net LTV. So in this case, we will also insert into the calculation some expenses. Now, when it comes to the question of okay, I have all of these, but what are the features? What are these things that helped me predict the LTV.
[07:27] Oren Cohen So here we are talking basically, for the engagement, onboarding quiz or questionnaire, lifecycle marketing, interaction with your clients or users, either email, newsletter, SMS, and obviously, transaction attribution and enrichment in case you do that. So this is all of the engagement for zero and first party data. And in some cases of companies that are also enriching their data. So these are all the ingredients if you want that you need to insert in order to have a proper prediction for LTV. I will say that, usually when you are looking at LTV prediction, it's not a constant or freeze picture of something. It's an ongoing process.
[08:11] Oren Cohen So you want to look also at some of the aspect of momentum, right what we call second derivative. So we want to look at product usage momentum, or pace of new users, or pace of features adoption, especially by the way, if you're looking at B2B collaborative companies, right when you have the single user, but you also have the organisation or the workspace. So all of these things that are related to pace of engagement and adoption, crucial momentum indicators for you to not only know what's your LTV, but to be able to act upon it.
[08:47] Shahin Hoda Got it. I want to dive a little bit deeper into an example of SAS possible, especially around you talked about the pace, the LTV pace is also really crucial. And you talked about even in the B2B space, can we can we dive into a specific example that you know, either maybe you can make it up or or you can think of Does anything come to mind so that we could we could dive a little bit deeper into this?
[09:13] Oren Cohen Yeah, absolutely. So so my apologies in advance for it. For going some into mathematics or calculation.
[09:21] Shahin Hoda Oh, we got into the meat and potatoes pretty quickly.
[09:27] Oren Cohen So one of the most challenging question is basically how you assess if you have a good LTV or not, right? So first, you need to make sure you understand the business case. So let's first understand why are we or what for that we are discussing LTV, and there are different models for different use cases, obviously. So if we are trying to give $1 value, let's say for every user and try to classify these user, if it's a one of the high value users or it's a low value. So what are we trying to do is we are trying to make sure that we are minimising the false positive, and that we are being able to bucket users according to their LTV. So let's look at the classic methods of many companies in the market today.
[10:14] Oren Cohen So they will look at the root mean square L, right? So I want to know, what's the standard deviation of the L of my prediction? In simple words, how often does my LTV calculation wrong and by how much? So if we'll take the very simplified example, that a sign up, let's assume will cost $50. So the 50 bucks is the CAC for signup. And for subscription, we'll assume a CAC of let's say 2000. So you build up a model of predicting an LTV, and you look at the evaluation of retrospective, and you see that the standard deviation is, let's say, 700. Is it a good or a bad thing? For which use case? And based on which base population? So if it's for signup, and we predict 700, it's very bad, right? So extremely bad. But
[11:11] Shahin Hoda You're in deep, deep trouble there.
[11:14] Oren Cohen Yeah. And if it's for subscribers, so 35% is not bad, given the fact that the standard deviation might be 3000, or 4000. So looking at the spread of variance of your subscribers' LTV. So what's really important here is how you calculate or how you build this model of prediction business wide. And there is no one LTV, it's really about the specific use case, we want to maximize the business impact.
[11:44] Shahin Hoda I want on pause, I want to interject right there, right? Yeah. So and just because some of our listeners might not have kind of statistics background. I just want to summarise what you talked about you tell me if I got this right. So you're saying we're talking about calculating LTV, and you said, hey, we might have a kind of sign up of $50 LTV, meaning that the majority, we took the average, and the middle of the bell curve is $50 for people who sign up.
[12:18] Shahin Hoda And then for subscription, we did the same thing. And right in the middle, if we again, create a bell curve of like how much each user is paying us, or sorry, how much each user is costing us is like $2,000. And then there is a deviation, meaning that most of the people who get to that stage, you said 700, so they're between 1300-2,700. So that's where that's the traditional calculation of LTV. Is that correct? Did I kind of summarise that right so far?
[12:51] Oren Cohen Yeah, exactly. And just maybe to to emphasise, so if your standard deviation, if your places in which you're seeing the O will be 700, and your CAC is 50, right? So we are talking about a very problematic situation. But if we were looking at the same standard deviation for a much cost, or something that is costing you more like 2000.
[13:17] Oren Cohen So obviously, this is not so bad, right? Here, suddenly, I can predict something which is in the ranges of of the price. This is exactly the question of it's not only about, yes, we are, let's say evaluating our model, according to standard deviation. This is also the question of for which base population and this is just one nuance which can shed some light on the on the difficulties or questions that's raised from you trying to utilise an LTV model?
[13:44] Shahin Hoda Yeah, because you might look at this number from, from, let's say, in the b2b space, you might look at it from an industry perspective, you might look at it from just like you said, from a signup or subscription perspective. So there are multiple different dimensions to look at this data. And am I right on that front?
[14:03] Oren Cohen Exactly. This is exactly the discussion that many times in Voyantis we're facing that the company will say, yeah, we have an LTV model. But then the question is which LTV model for which use case? And do you see it effective during the actionable timeframe that you have with networks or other lifecycle marketing?
[14:25] Shahin Hoda Okay, got it. So this is a traditional way of looking at it, right? What are some of the other methods that you would look at LTV and calculating it?
[14:35] Oren Cohen Yeah, so I think it's worse to just try and nail the differences. So let's say that you want to have your prediction LTV for the sake of bid adjustment. So you optimise for signups, let's say because you don't have enough events for subscribers, so you're taking an event which is more frequent. And you need a model that will predict the LTV for a cohort of signups user within the conversion window of a given network. Doesn't really matter. Another example is if you want to prioritise salespeople for upsells to enterprise accounts, right?
[15:08] Oren Cohen We all know that many of these B2B Sas are driven by enterprise sales team. And the question is, if I have only one phone call to do, what is the hottest lead that I can approach. So you need basically a team LTV model of a workspace LTV model, which might apply even month after signup. So here we are seeing the use of two models, one must be very immediate, because you have the limitation of the attribution window. And this is the case of LTV prediction for feeding user acquisition channels, while the other or the second use case is talking about feeding salespeople without the limitations.
[15:45] Oren Cohen So this is just an example of how the use case can affect the exactly what will be the structure and models that you're going to be. Now this is why I'm always saying that LTV is a very smart concept, but you really need to be in the details of what is that goal that you're trying to achieve. And this will let you walk you through what is the base population and devaluation method.
[16:10] Shahin Hoda Got it. Okay, let's talk about, let's change gears a little bit and talk about CAC. And I know you have a love and hate relationship with CAC, where traditionally there's been a lot of emphasis put on CAC. I remember some time ago, a client of ours turned to me and said, hey, you know, what is a good CAC? Like, we did the calculation, I can't remember the number, but let's say their CAC was quite high, you kno. We did the we did the calculation, and our CAC was, you know, $30,000. Is that a good number? Is that a bad number? And I kind of turned around and said, hey, it really depends. You can't really answer that question. So where does CAC come into the equation?
[16:55] Oren Cohen Okay, so this is probably one of the most spoken subjects in every call that we are having, regardless if it's B2B PLG or even a B2C subscription company, right? So, I will try to navigate this in the following. So cash flow is your limitation, right? We'll discuss it and payback period is your goal. Now this is what I want to, this exactly my statement that CAC is not a goal. So what your control: you control the date, you control the budget, and the channel allocation. In CAC is basically all costs of acquisition is the outcome of how you allocate your budget, the competition and your optimisation technique.
[17:38] Oren Cohen So far, we see many companies aiming for CAC target without taking LTV into account. And these days, even measuring the CAC is not an easy test, given the privacy changes. Now, just from my personal, let's say, if I'm looking at the market, even the simple news about Netflix will always say how much new subscribers Netflix added in what was the cost of acquisition. So this is why this mindset is set for all of us because this is something that the public markets are using, but there is no one CAC, right? At the end, there is a CAC for signup, or a CAC for a subscription, and you really need to navigate.
[18:20] Oren Cohen So we should remember that sometimes CAC is just a measured metric under the constraint of attribution window, the time to act. However, there is another event in most of the, let's say, especially PLG, but the B2B SAS companies, there is a different event, which is much deeper in the funnel, which is really correlates with your success and revenue generation. So it might be the case in which with the same CAC, let's say, the CAC for signup, you manage to lower the cost pair, your aha moment. This specific deeper into the funnel event that is really correlated with revenue and LTV.
[19:01] Oren Cohen So I think that CAC is a great metric you should look at as the outcome of your allocation, the way you compete with others, and the way that you optimise your campaign, but you need really need to look at what's the cost of this aha moment, this moment that you know, this workspace is going to convert, this organisation is going to reach out sales. So CAC is part of the equation I said, but it's more of an outcome than a North Star for any company.
[19:30] Oren Cohen And our obvious recommendation will be to look at the LTV to CAC ratio. So if the CAC, if you have a network and you see that the CAC is double, but you see that the predicted LTV is 10x, probably you will be happy with it. But I guess first of all the company need to know how to measure LTV, then they need to know how to predict the LTV and then this ratio of LTV to CAC discussion can be much more informative and cascade across the different departs percent of the company.
[20:01] Shahin Hoda Okay, so correct me if I'm wrong, you're saying instead of just looking at CAC, as a, you know, this random number that could be, and I love the quote that you said there, there is no one CAC, there are different CACs. And what you really recommend is look at the. am I getting this correctly that you were saying? You got to look at the CAC, the customer acquisition cost of the aha moment. See how much it costs you to get someone to the aha moment, which is a lot closer attributed to the bottom line revenue. Is that correct? That was the first thing that I want to bring up, yeah?
[20:40] Oren Cohen Absolutely correct. And it's with a strong emphasis when it comes to B2B PLG, SaaS companies that have a very long lifecycle, they have a freemium model. You can utilise the product without paying a single penny for very long time. So it's not about how many signups they're getting right. It's what about how subscribed, how many subscribers they're getting. And these subscribers will usually have some kind of aha moment that from their own, you can be quiet and say, okay, this workspace is going to convert.
[21:11] Oren Cohen So this is exactly what I'm talking about. And keep in mind that most of the time, you're buying over your user acquisition based over network like Facebook and Google. And they required from you to work in optimized dollars in events who have an was frequent enough, right? So you have tons of signups and therefore it's great for optimization. But this is not the real outcome, or the real aha moment that you want to measure. So this is exactly what I'm trying to say here that you need to look at the cost of these aha moments, not necessarily the classic definition.
[21:47] Shahin Hoda Got it. And then the second thing that you brought up is, what's also important is the LTV to CAC ratio. And this is a great point, because in the example that I talked about before, hey, our CAC customer acquisition costs $30,000. Well, doesn't matter how much is your LTV, like what is the LTV that we're looking at? If the LTV is a million dollars? Well, that's nothing that's just peanuts. Until totally makes sense. What what is kind of the rule of thumb advice that you give to people about what kind of ratio should they be looking at? What is that range look like? What is maybe the minimum? I don't know if there is anything that you can kind of provide as a guideline to some of the listeners?
[22:28] Oren Cohen Yes. So you know, in our market, and especially around DTC companies, so everyone would say three to one. So I'm not, I'm not saying that there is a plug number one magic number or the magic ratio, you should aim to and just want to nail the point that first of all, you need to understand what is your predicted LTV, and hence, you can understand what is the, let's say, wide range of payback times that you can use. So obviously, if you're just a new startup, and you're trying to raise money, you cannot live with a CAC ratio, which won't meet your cash flow.
[23:04] Oren Cohen So it's really depends on what is the strategy of the company, whether they are the growth mode, and they just need to show higher numbers of signups or they really as the market today, they're really being expected to meet profitability. And then this is a different answer. So I won't say there is one plug number, this is an answer that probably a VC guy can can come up with, for me to really know your data, and then understand what you can do with the limitation that you have.
[23:32] Shahin Hoda Okay, let's talk about data. It's a great segue. A lot of these things that we've been talking about, man, it's data hungry, like it needs a lot of data points. How can organizations improve the data that they have that they're feeding into their LTV model? What are some of the advice that you have? They're
[23:55] Oren Cohen Great. So I think you know, first of all, it's like the given end result of every startup or a big company that there is this moment in time, when you understand that your data is not sorry, then you don't have the well structured data team. And so it's really about first of all, knowing your data, make sure that your data is in its real form. But let's understand how a company can increase or affect the amount of data that they have. Let's take the simple, most simplified example of "ask your users."
[24:26] Oren Cohen So there are a lot of cases in which if you have a smart, detailed onboarding process, which have a lot of quiz, questionnaire, a lot of indicators that the user can just say, what are their behavior usages, how frequent for how many people, why they use the product. These things are the zero party data that the company can get in order to later feed this as features into the predicted predictive LTV model? So let's take the most simple example. Let's say that this is a B2C company and it's a food delivery app.
[25:05] Oren Cohen If you ask users for how many people you order food, what's your favorite dish? And how many times you order a week? Probably, if it's seven days for five people noodles, this is a high value users. So this is just to give you some kind of an understanding of how zero party data of these onboarding funds and processes the product usage question can feed with the data. And obviously, there is also the zero party data, which is really mapping and making sure that you measure every single interaction within your website, have a product or platform.
[25:40] Oren Cohen And this is exactly what we recommend. There are obviously cases in which also filled out enrichment. So just working with one of these companies that will enrich your data based on what they have. But I will say that in these areas of data, always try to have a sufficient zeal and feels about the data and Only later did with enrichment if to say so.
[26:02] Shahin Hoda Alright, can I ask you a basic question for some of the listeners who might not know because you've brought up the concept of zero? First party and third party data? Can you define the two and differentiate the two for for our audience?
[26:16] Oren Cohen Yeah, so I think here, the question is really whether this is, let's say, quote, unquote, a side effect that you have this data because you measure it and someone engaged with your platform or product, or whether this is data that you actually directly extract the intent from the user. So everything which involves KYC, onboarding, quiz or questionnaire, a funnel with the product selection to tailor or custom the product, especially for you.
[26:45] Oren Cohen All of these places in which you directly ask the user for something. And it's not the outcome of the usage, this will be defined as zero-party data. And first party data is obviously all the data that you have, as a first party, because clients or customers organization are using and engaging with your product or platform.
[27:07] Shahin Hoda Thank you very much for that clarity. I want to talk about obstacles. What are what are some of the obstacles, and we touched on some of these already. But let's dive a little bit deeper in terms of what are some of the obstacles that prevent companies from utilizing or building LTV models?
[27:23] Oren Cohen Right. So first of all, I think let's understand that. The question, again, is for what you're building this LTV model, so and the question is who is building the LTV model? So in case that the data science team that usually deals with the product, trying to build an LTV model for the sake of user acquisition, this is something that you need to understand whether the data science team has the understanding of what's the business use case.
[27:49] Oren Cohen But let me maybe list view. So I think many times there is no evaluation framework. So you're testing and you're trying to build an LTV model, but you find yourself with missing the knowledge of how to evaluate the model, whether it's prebiotic or constant evaluation. A lot of time what we see is oversimplification. So you see quick and dirty Max business impact. So factors assumption, I assume that all of my cohorts will behave the same, and my revenue will always be x in day 20, or will z into day 180.
[28:27] Oren Cohen So this is trying to simplifying and probably you will miss a lot of the efficiency. The assumption that your LTV same across clients. So many will say my average LTV is stable when I'm looking a few months. So the average might be stable, but you think you don't have big variance in the LTV across users. And while average LTV might be the same, however, it holds a big variance among users are workspaces. So the misefficiency is that you end up paying the same price for high value users organization and low value users. So this is exactly some kind of the the thoughts or the mistakes around LTV.
[29:12] Oren Cohen Now just want to if I may have an example, that will address your last two questions. So what are the obstacles and also how data can be used? So let's look at a very simple example of two ad groups with zero subscribers and 1000 signups. So every user acquisition almost we look at these two ad groups, and we'll say they are both bad and they are the same. However, you have now the data of 1000 signups that you can analyse from all of the aspects that I described as zero and first party data. And there is a case in which one of the ad group generate highly engaged signups while the other ad group generated a low engage signups.
[30:01] Oren Cohen So instead of looking at I have zero conversion for subscribe for subscription, you need to look at the 1000 signups that you have and try to extract for New Zealand first party data if these two groups are the same, this is exactly the type of input that you can get from LTV prediction, if you're looking at the bass population of signups and the fact that you haven't managed to convert any of these to inactive subscribers, by the way, you can do the same with predictive intent model.
[30:31] Oren Cohen So we are discussing here, the LTV, the lifetime value, but there is also the intent whether this workspace so this is a client is going to convert or not. So this is just a simple example of how companies they their mindset is to look at, okay, I don't have conversion, but they are missing the 1000 users per ad group that they have in their data lake or data warehouse, that they can analyze their behavior and come up with a predicted LTV. So this is a classic example of the mindset that blocking you for implementing or utilizing LTV model, and the data that you have in hand and you can start utilizing today.
[31:10] Shahin Hoda I love it. I love it. Oren, I got I got one last question that I want to ask. I mean, there might be more questions that come out of it. But one main main question that that I want to cover, let's say I'm a marketer, or you know, I'm in the revenue team, or I'm in the analytics team, and I'm sitting down, and I'm like, okay, you know, our, our LTV calculation is definitely wrong, or they're definitely gaps in how we went about it.
[31:36] Shahin Hoda But I feel a little bit overwhelmed listening to Oren because it's just like, all these things I got to take into consideration and this, it just feels like I'm going to be trying to lift the world here to make this happen. Where should I start? What should be the starting point for for someone like that, and I will take out the easy answer. You can't say they gotta go and buy Vantis to this. So you know, where? Where should I start on that?
[32:09] Oren Cohen So first of all, I like your suggested idea that probably this is not the answer that you're aiming for. So I will say the following is really starts with the data. If you don't have sorted enough row form data, which is not immutable, and that you can trust, this is where you need to start. But I think this is not serving the sole purpose of LTV prediction, this is serving the whole growth loop of every product. We know that data is crucial for you being able to see reality and therefore make educated decision.
[32:39] Oren Cohen But I will say that it's really after you understand that you have the data. The question is, where's the place within your business that you have the biggest pain point, if sales is the issue, probably you need to understand how to deal with subscribers, that should get the attention of salespeople. And there, it's not really Mather about other segments, you want to focus only on these subscribers, maybe because you want to upsell to an enterprise account.
[33:08] Oren Cohen And the second thing, it's not really important, whether what will be the predicted value of dollar sign or revenue tag of each one of these, you just want to index them and sort them by the probability to convert. So this is just an example, right? But let's take a different example, which is extremely common in the B2B SAS world, you want to support user acquisition. So you are coming to the user acquisition team, you're asking, we want to build an LTV model or an intent predictive model that will help you and then probably you will hear from them.
[33:40] Oren Cohen We don't have enough events to optimise our campaigns towards subscribers. So we are using signups. And then it's obvious for you that you need to come up with an event that holds 100% of the subscribers. But the top signups as well. In here, you have an event, which has enough free which is frequent enough, in order for you to optimise towards you have your subscribers that it's the base population, but you're also adding let's assume 20% of your top signups in here you have an event which has extreme amount of data and that you can optimise.
[34:18] Oren Cohen So I wouldn't try to have an LTV or intent model, which is covering all of your pain points, I would focus on one thing points understand what are the requirements, whether I need to be accurate, I need to minimise false positive, or I need to increase the amount of event. And from there, I will look at the data that I have in order to address this. Obviously, very tough.
[34:41] Oren Cohen If you're a company with a DNA of data science serving product and suddenly you're giving them the extra effort of serving the user acquisition team and they don't really familiar with the concept and they don't know whether this module is going to be used. So I will say that if you have the resources and we're assuming that you have otherwise, you need to come to us try to find this very immediate pain point well defined that you can address utilising your own data.
[34:10] Shahin Hoda I love it. I love it. So, right. So define your objective, start with collecting the data, define the event that you want to measure. And then and then mix. And if you realise that you don't have enough kind of interactions on that event, mix the data that makes sense so that you get to statistical significance and start from there.
[35:30] Oren Cohen Exactly. I think it's a common challenge for B2B SaaS companies that the lack of event the early days of the use of Germany because of the freemium model, and this is, you know, the, let's say, predicting the intent in knowing today how to act for these ad groups, these campaigns, these channels is the one of the secret sauces of most of the biggest b2b freemium companies that rule. The recent years, I would say,
[35:56] Shahin Hoda Oren, before we wrap this up, is there anything else that maybe I didn't cover? You think it's important for us to talk about on this topic?
[36:05] Oren Cohen Yeah, I think just maybe to look at the LTV concept. From a different mindset. I completely one of these people that don't like to hear a lot of buzzwords around the industry. But I think, given the need for profitability, the recession in the markets, and the privacy era, which we are now facing, it's really important to make sure that there is someone within the organisation that understand the concept of LTV, and more other than this, if you're using LTV for the sake of improving your Google Facebook campaign, you need to understand that there is the know-how of data, they know how data science, they know how of growth or user acquisition, and know how of how to work with these network.
[36:50] Oren Cohen And you need someone who will master all of these. And every journey starts with a step right. So at least don't look at LTV and say, this is another concept that everyone talks about that no one really gives, right. So I would say, start to understand where your company today when it comes to LTV prediction, and don't wait for the time in which most of your competitors will do something like that. And you will suffer from the MIS efficiency, just because everyone optimising don't future value value. And you're still stuck with Hiscock mindset. So this is, let's say my advice to folks who listen to us.
[37:28] Shahin Hoda I love it. Let's do some rapid fire questions before we wrap up or let's do it. First question I want to ask you is was what is one resource? It could be a book, a blog, a podcast, a talk, whatever it is, that has fundamentally changed the way you do the work or you live but has it had a very profound impact on you? What comes to mind?
[37:48] Oren Cohen Yeah, so I will say a Man's Search for Meaning by Viktor Frankl. So I think this is probably the book that most influence how I look at life. So, Friedrich Nietzsche once said, who has a why to live can bear almost anyhow. And I think this book just led me to this understanding with the right why I'm doing things I can do act in every environment or situation.
[38:16] Shahin Hoda I love it. That's definitely that's definitely a classic. Alright, question number two is, if you could give one advice to B2B marketers, what would it be?
[38:24] Oren Cohen So I would say start using prediction models for prediction of the LTV, the intent, sales scoring, whatever, but start playing. Start testing. Start engaging with this, because they think it's going to be one of the secret sauces for successful company at following yours.
[38:43] Shahin Hoda Love it. Love it. Question number three, who are some of the influencers that you follow?
[38:48] Oren Cohen Yeah, so I spent 10 years at Iron Source. So all of the mobile space and especially mobile games where the fields of expertise, so I really like Mobile Dev Memo by Eric Shepherd. So I think this is probably a blog, which not only keep you up to date with the news in what's going on in the industry. It's also a blog in which you can find tremendously deep, detailed explanation of most of the most important common growth aspects. So this is really a place which I read something and then saying, Okay, I learned something new. And I really recommend to give it a look.
[39:27] Shahin Hoda Last thing, I want to ask is what's something that excites you about B2B today?
[39:31] Oren Cohen So, the team knows that working with them, there were notion just made me extremely excited about these collaborative B2B PLG companies. The fact that you have business which hold tons of growth loop, the challenge around the private usage and the workspaces, the mobile funnel, which is very hard for these PLG B2B SaaS, the long payback time, tons of use cases.
[39:59] Oren Cohen So we know that Miro or Notion can serve you in so many verticals or use cases. So I think these companies that managed to base their business on cast cascading into workspaces and teams using tons of growth loops that feeds each other these are probably the rising stars that I really, like, a geek, looking at these and reading everything, because I really think this is the way to go.
[40:24] Shahin Hoda I love it. I know it's amazing that you guys are working with them. And kudos to you and the team for making that happen. So and, you know, and again, thanks a lot for coming on the podcast. This was an insight rich, packed podcast that I think if anyone's listening, they probably have taken a lot of notes just like I have. Thanks a lot for coming on the on the podcast, Oren.
[40:49] Oren Cohen Sure anyday, and it was extremely fun, and I hope that I managed to ring some bells for people otherwise, I just heard their time. So
[40:58] Shahin Hoda I'm pretty sure you have. I'm pretty sure you have and I'm looking forward to future conversations.
[41:05] Oren Cohen Thank you so much.