---

Jacob on LinkedIn: https://www.linkedin.com/in/jacob-rushfinn/

Botsi: https://www.botsi.com/

retention.blog: https://retention.blog

Price Power podcast: https://podcasts.apple.com/ng/podcast/price-power/id1844494362

Newsletter: https://intelligentartifice.kit.com

FULL TRANSCRIPT BELOW

Shamanth: Welcome to Intelligent Artifice, a part of the Mobile UA show.

Shamanth: Intelligent Artifice is a deep dive on how AI is transforming growth and performance marketing.

Shamanth: Since 2018, over 250 plus episodes, we’ve been deconstructing how the best growth marketers actually win.

Shamanth: And every week, I sit down with one operator shaping the future of growth in an AI world.

Shamanth: My guest today is Jacob Rushfinn.

Shamanth: Jacob is the CEO and co-founder of Botsi, a platform that personalizes the offer each new subscription app user sees.

Shamanth: He also writes about retention and monetization at retention.blog and hosts the Price Power podcast.

Shamanth: I thought this would make for a great episode because most teams still treat price as a single number to AB test.

Shamanth: Jacob’s view is that price is only one variable inside the whole offer, and that there is no one right offer for every user.

Shamanth: In this episode, Jacob walks us through his method for price testing and the signals he uses before the paywall to personalize an offer.

Shamanth: He explains why, for some high-intent users, a lower price actually can double conversion.

Shamanth: He shares a check any team can run today without a machine learning model.

Shamanth: We also get into how personalized offers work within Apple’s rules.

Shamanth: There’s so much gold here around monetization and testing, and I’m excited to share this with you. Let’s get into it.

Shamanth: I’m excited to welcome Jacob Rushfinn to Intelligent Artifice. Jacob, welcome to the show.

Jacob: Hey Shamanth, super excited to be here. Yeah, thanks for having me on.

Shamanth: Yeah, I’m excited to have you know, because I’ve followed your work for a very, very long time.

Shamanth: And followed your blog a lot and I think you write incredibly in-depth analysis of apps, products, on-boardings, everything. I feel like I’ve learned

Shamanth: Stolen a number of ideas from that, which has been great. And you also run very an incredible podcast, which I was fortunate to be on.

Shamanth: And you are one of the thinkers I look up to for all things pricing and monetization.

Shamanth: So, thank you again for making the time for this.

Shamanth: And you know, a good place to start really would be,

Shamanth: You know, you wrote about you know, five-step method for price testing.

Shamanth: Can you talk to me about, you know,

Shamanth: How first, of course, talk to us about the method itself.

Shamanth: And then you can I would also love for you to talk about

Shamanth: How you approach price testing when you were in-house,

Shamanth: Which is a couple of years ago, versus how that has changed post AI.

Jacob: Yeah, it’s a good question and price is such a big lever for your app, pricing and product and kind of packaging holistically.

Jacob: I think when we talk about price testing, we really don’t want to limit it to specifically the price.

Jacob: We want to think about the offer as a whole and everything someone is seeing where the price, the product, the trial, how it’s presented is all also connected, the other products that are presented with it.

Jacob: And so we want to start thinking about how we prioritize which elements to test first and which to optimize.

Jacob: And so a lot of this comes from Michael Perizek’s blog post from his testing while he was at Mojo.

Jacob: And so what he found and what he suggested was that like first we need to figure out the baseline price.

Jacob: What is this core initial price for your main market?

Jacob: And then this is, you know, standard AB testing. We can get more into the kind of personalized approaches, but you know, what they did was, yeah, AB test that first baseline price.

Jacob: And then go for your other larger markets,

Jacob: AB test those prices as well. So let’s say, you know, your biggest market is the UK and your, you know, next couple biggest markets are, you know, France, Spain, and the US.

Jacob: Okay, well, you can’t AB test every single market individually, but you want to get those base kind of local prices for your biggest markets right.

Jacob: Okay, now you have figured out, and this may take a few iterative AB tests.

Jacob: You can launch, you can speed this up a little bit depending on your volume by launching a higher and lower price than what you have today

Jacob: In instead of just, you know, two variants, you can have three variants. This will let you get a read whether higher or lower is performing better and try to help you understand kind of renewal rates and cancellation rates from that.

Jacob: Then you want to think about the packaging. And so what is packaging?

Jacob: Packaging is everything on the product that isn’t price. So it’s the often the subscription length, but also the different products you offer.

Jacob: And so, you know, for most subscription apps, you’re going to want to try to get more people on an annual plan where this is higher LTV and more value.

Jacob: And so think about the ratio between maybe a monthly and annual plan or a weekly and annual plan or quarterly and how this is influences purchasing decisions.

Jacob: You know, figuring out this kind of right comparison, how much is the annual discounted.

Jacob: And then also, you know, so it’s kind of that ratio, the number of products, and the subscription length of those products.

Jacob: It’s kind of this first packaging step. And those are pretty big levers too if you get those right. They can really move the needle.

Shamanth: Yeah.

Jacob: Where, you know, potentially raising your monthly price can, you know, have a huge impact on RPU and LTV because you’re driving meaningful more subscribers to annual plans.

Shamanth: Right. Because there’s the anchoring, right? Because they use us to compare the monthly to the annual and they’re like, oops, this is expensive. Let’s just do the annual plan.

Shamanth: Interesting. Yeah. So yeah.

Shamanth: And you know,

Shamanth: And a lot of your work and approach right now is around personalizing price and let’s get to that.

Shamanth: Can you talk to me about

Shamanth: What the was there an inflection point when you were like,

Shamanth: Showing one price for everyone just feels wrong, even if you AB test this, right? So you could AB test everything and still show the same price to everybody.

Shamanth: So was there a point of realization where you’re like,

Shamanth: This is not optimal.

Jacob: Yeah, it started with kind of this price testing process that I was talking through.

Jacob: And I think a lot of people have feelings around this when you start trying to localize for different countries,

Jacob: Where it’s just a such a difficult process of trying to AB test country by country. Maybe you group some countries in the Euro zone together, so it’s a little faster, but and then ultimately you have higher intent, lower intent countries that maybe don’t have enough volume to AB test any reasonable amount of time.

Jacob: And this country isn’t the only signal you should be using. So you start kind of doing all this manual work and it’s very slow and

Jacob: And then you also kind of maybe you raise your prices and you start seeing that, okay, we made some more revenue, you know, up front, this is going up, but then like our total subscribers dropped and, you know, we’re a subscription business and so this is supposed to be built on compounding and stacking cohorts over time that are renewing and staying subscribed.

Jacob: And so it you start to think about, okay, well, there’s different willingness to pay and there’s different intents under the hood here.

Jacob: What are these signals I could use? And I think a lot of people start with more targeted rule-based approaches where you can see that maybe based on an onboarding question, someone’s specific goal shows they, you know, they have a much lower intent and you’re going to show that user a discount.

Jacob: This is often how people start, but then once you start laying on more complexity, you realize this is very the rule-based approach is very hard to maintain and it doesn’t kind of adjust for things in real time.

Jacob: So these are these early signals around, hey, like there’s got to be a better way to do this.

Shamanth: Yeah.

Shamanth: And what is that better way according to your current approach with Botsi?

Shamanth: In other words, you know, what signals does your platform account for when figuring out, look, here’s sort of the kind of price point I need to show this user.

Jacob: Yeah.

Jacob: So there’s a few components here. One,

Jacob: With you know, with other businesses, you have a much longer tail of conversion

Jacob: In the sense that people can use products for a while and then pay over time.

Jacob: And you kind of maybe see this with e-commerce or retail or even games, right?

Jacob: But with subscription app businesses, it’s all about really that first conversion.

Jacob: You need to convert someone on day zero. You know, there are a few exceptions, but, you know, 80, 90% of businesses, if you don’t convert them on that first session, they’re gone and never coming back.

Jacob: And so really figuring out that first offer is the really critical one. And I say this up front because that dictates the signals that you can use to personalize these offers.

Jacob: And so what we’re doing and what we started with kind of, you know, beta testing and private clients last year and now we’re, you know, doing for apps that are doing more than 100 million ARR and so it’s pretty exciting to kind of see the results is looking at all the signals we can get pre-paywall during this onboarding.

Jacob: And so we look at things like, you can get all the device data. This is things like, you know, country, OS, device type, language.

Jacob: There’s some other different attributes in there as well. And then there’s the first party data that someone may have you know, through a quiz questionnaire flow.

Jacob: You know, not every app has this, but often there’s some additional first party data you can get.

Jacob: And then also if we can layer in acquisition attribution source, so can we get ad network, campaign ID, you know, on the Apple search ad side, you can get super granular in terms of keyword.

Jacob: Words, and then, you know, you try to get some probabilistic matching to understand and try to start associating maybe ad sets or ads or creatives as well, and using these as a signal. And of course, this varies by network, varies by app, by platform. But this is these is the core of kind of these three layers of device, first party data, and then acquisition, attribution source.

Shamanth: Yeah.

Jacob: And this is important, but you also want to think about

Jacob: You have to kind of layer this. You can get with this, you can get stuck in conversion optimization, and specifically trying to convert more users, right? But ultimately, you have to layer in this predictive LTV signal as well. And so you’re actually truly optimizing for, you know, long-term value and revenue versus just kind of that conversion.

Jacob: And so combining this these early signals with this longer-term kind of predictive revenue is really how you lead you get a very very strong result.

Shamanth: Yeah.

Shamanth: So if I understand correctly, you look at the downstream long-term LTV of past cohorts and you’re like, hey, this is a high-end iPhone from San Francisco, and you know, coming in from XYZ campaign,

Shamanth: And you know their LTV is high in the past, and therefore you’re like, let’s show a high price to this user. Is that roughly the approach?

Jacob: Sometimes, but I wouldn’t

Jacob: I wouldn’t get caught up in the price. Price is just one variable.

Shamanth: Right.

Jacob: It’s holistically the offer.

Shamanth: Right.

Jacob: And so we also want to think about really what we’re doing is more of a kind of online training reinforcement learning process where historical data is important, but

Jacob: If you were just looking at historical data and this was predictive of the future, our lives would be a lot easier, right? Like things are constantly changing. If we could just kind of look at the past year, but your product is changing, the competitors are changing, your acquisition world is changing. And so really it’s an evolving signal that is more front-weighted in terms of and so figuring out how to weight the kind of time decay of your data signal is pretty important.

Jacob: And then it’s figuring out how do different offers affect this LTV. And so a lot of times it’s taking these historical baselines, building out your renewal and retention curves as a starting point, but then you have to dynamically adjust them based on the product.

Jacob: And so, right, if a user is seeing a higher value product is buying monthly versus annual with a discount, this is going to change that renewal curve. And so the trick is how do you figure out how the renewal rates are going to change within the first few days to a week, and so you have a fast feedback loop.

Jacob: And so what actually other people can do today, which I recommend is that you don’t need a machine learning model to do this, but look at your cancellation rates after a user has subscribed and converted,

Jacob: Figure out the percentage of users canceling in the first seven days to 14 days, you know, post conversion, and this ratio of cancellors is going to be highly predictive of cohort quality in terms of the end renewal rate.

Jacob: So you can actually use this the percentage of people canceling in the first 7 to 14 days as a predictor of renewals at one year.

Shamanth: Right.

Jacob: And you can I mean for you can use this for monthly, it’s even easier for monthly because this follows a more steady curve, but this is and you want to layer this in with an engagement and other signals as well, but most people are very bad at simple estimations and when you’re they’re running these AB tests of different offers, they’re treating like everything is the same, that things are going to renew at the same rates, but really, you know, it’s a dynamic world and so how we want to think about this is it’s a kind of learning reinforcement model that’s testing and exploring.

Jacob: And so we when you’re only dealing with historical data, it’s hard to move to a causal thing of actually proving cause and effect. And so this is what the models are able to do with kind of experimentation, testing, learning, and then this continual optimization process.

Jacob: So, a long-winded answer, but hope hopefully that explains some details.

Shamanth: Yeah. Yeah, right, because I think it’s very easy to just optimize for the first transaction and as you just said, make it conversion rate optimization, which if you’re very small, if you’ve never done testing, is probably a low hanging fruit, but if you’re at substantial scale, if you’re just already tested a ton, then you’re like looking at, look, how does my retention curve change over 1, 3, 6, 12, 15 months,

Shamanth: And how could you just predict it sooner, right, based on the offer, but also based on the user characteristics, if I understand correctly.

Jacob: Yeah, and I think this is something smaller or medium apps, you know,

Jacob: Start if you’re doing kind of doing this yourself, start with just like the high-level cohorts by product. The bigger you get, the deeper you want to segment. And so yeah, we you know, we’re kind of having these renewal curves and predictions by, you know, as many different kind of segments and cohorts as possible to kind of get as accurate.

Jacob: And then we kind of, you know, combine these all for this kind of blended rate, but we also have these insights and so it’s super interesting kind of like what different signals predict what and you know, what one thing I you know, you mentioned earlier like, you know, showing a higher price to higher converting users.

Jacob: You know, actually, sometimes it’s the opposite. Sometimes we’ve seen that the highest intent segments, we can actually meaningfully improve revenue by showing a lower priced plan or lower priced offer because the intent is so high that we’ve double conversion rates by showing a lower price.

Jacob: And so it’s an interesting where every product is different and this is you could never understand this by just modeling out historical data and just going, oh, show high converters, higher prices.

Jacob: Really, like you’re your user bases are so unique and so dynamic that you really need to have kind of this experimentation and continual learning process to understand what actually is the right offer for these different segments and groups.

Shamanth: Right. And if I also understand correctly, because you need to understand retention curves by very different dimensions and by very different offers, you need a certain critical mass of users to run an experimentation program of the sort you’re talking about.

Shamanth: So, is there like a benchmark or a critical mass sort of thingy?

Jacob: Yeah, we typically find apps doing more than like a million dollars a year, we’re able to show value for. And you know, revenue is kind of a proxy for data volume.

Jacob: It’s an easier thing because like, you know, different price points equal different kind of conversion rates and different number of converters.

Jacob: Ultimately, it comes down to the number of conversions that you get in a period of time.

Jacob: And so to give you a sense like we aim for 500 or a thousand conversions to kind of launch the first iteration of the model.

Jacob: After that, it kind of goes into kind of continual learning.

Shamanth: Right.

Jacob: But we want to get there. And so this typically and if we typically want to get this in a couple weeks, right?

Jacob: If you can’t get that volume in a few weeks, it just means the improvement cycle is going to be too slow.

Shamanth: Correct.

Jacob: And you know, personalized offers can still show some value for you, but the it’s not going to be a like super transformative experience just because the waiting time and lead time is going to be a bit long.

Shamanth: Yeah. So, and also, the way you’re also describing your approach, correct me if I’m wrong, but a lot of this would still have been possible pre AI, using machine learning algorithms, and if you had enough engineers, you could build something like this.

Shamanth: So, did AI change things substantively in your approach at all?

Jacob: Yeah, it’s a great question. So, what we have seen is that yeah, the a lot of the underlying pieces are machine learning, but previously it was really hard to build and manage custom models per client.

Jacob: And so we don’t have one generalized model that everybody uses. We actually train custom models for each client. And maintaining this, managing this, kind of training and building these different models for each client,

Jacob: We’ve been able to do successfully with the support of AI. And so we have kind of a AI helper layer.

Jacob: And so, AI that in terms of LLMs is not fast enough or deterministic enough for what we’re doing.

Jacob: Where we need to have a response in 50 to 100 milliseconds,

Jacob: Or, you know, 1 to 200 milliseconds on kind of edge cases.

Jacob: And so this is, you know, LLMs, not really the right use case.

Jacob: Maybe in five years, but I still think the deterministic aspect is, an issue.

Jacob: And so really where we see the AI is this kind of helper supporter layer that allows us to kind of scale.

Jacob: You know, of course, all the software development side, you know, we’re, you know, building as well and maintaining.

Jacob: And then we’re right now we’re working on

Jacob: Adding in kind of future recommendations with an AI layer.

Jacob: And so, the more continuous this process can be in terms of, okay, you picked these first products, you figured out kind of what’s working.

Jacob: You have a good personalized offer model running.

Jacob: What’s next? What do you improve? What’s not working? What can you take away?

Jacob: And so the AI layer is super valuable to kind of quickly diagnose and dig into your data and offer suggestions.

Jacob: And so we’re working on kind of training specific, you know, recommendation engines based on kind of the subscription app world that’ll be helpful for people to kind of have this never-ending optimization process.

Shamanth: Interesting.

Shamanth: And, you know, these sort of custom models and this sort of personalization, given that this wasn’t anywhere on the horizon even a couple of years ago,

Shamanth: What did you personally have to learn in order to be able to

Shamanth: Understand the viability of this, but also to actually execute this.

Jacob: Yes, so, you know, I’ve I took some stats classes in you know, high school and college and so I can run a really good AB test, but I’m certainly not a data scientist.

Jacob: And so, you know, we were, you know, we had this idea of,

Jacob: There is not one right offer for every user.

Jacob: We should be able to show different offers to different people.

Jacob: And so it kind of started with what I was talking about before of,

Jacob: We had this intuition feeling.

Jacob: We started looking around and saw that, you know, some of this existed in terms of like, you know, Netflix’s kind of personalization algorithms to show, if you think about all the thumbnails for different content on Netflix, they’re personalized for you.

Jacob: They’re actually figuring out which ones you’re going to respond to and they’re optimizing these over time.

Jacob: And so different people see different you know, thumbnails for you know, movies or TV shows.

Jacob: And so it’s a similar concept here of, okay, how does that work?

Jacob: What is happening here?

Jacob: And then what you also think about ad networks.

Jacob: Ad networks are solving some type of similar problem where you’re trying to match the highest value ad with that user.

Jacob: What ad should you show to this person that they’ll convert with, but also you lay on this revenue component that’ll generate kind of revenue for you.

Jacob: And so this was a tangential problem as well.

Jacob: And so we started digging into how kind of all these things are working under the hood.

Jacob: You know, watching a lot of YouTube videos on different kind of ML models and we started to go, okay, like this can work.

Jacob: And then we you know, started working with some you know, great people that had had built out different kind of personalization engines and you know, kind of dynamic engines before in different industries and that’s really kind of how we kind of came to where we are today.

Shamanth: Wow. Yeah, and I think

Shamanth: It’s from the looking from the outside, it’s not very obvious as to how technically complex and challenging this can be.

Shamanth: And amazing for you that it seems like you actually had to learn and figure out a lot of this.

Shamanth: You know, even though you had the initial kernel of the idea that look, we just want to price personalize, but just the actual execution was a very, very heavy lift.

Shamanth: Amazing.

Shamanth: You know, the other layer of complexity which I want to ask you about is, you talked about, look, different prices for different people.

Shamanth: How does that work within Apple’s rules?

Shamanth: Because technically somebody would have to basically would they submit a hundred SKUs and then pull one of those on the fly?

Shamanth: How would that work?

Jacob: Yeah, so, it, you know, first off, yeah, completely compliant with Apple’s rules.

Jacob: The important thing, is that you can think of this as like

Jacob: Advanced kind of micro segmentation where there’s a limited set of products that are available.

Jacob: You’re pre-selecting, pre-defining those products.

Jacob: You’re making them available in your app.

Jacob: Apple is approving them, you know, through the submission process.

Jacob: You know, Botsi isn’t dynamically generating new price points on the fly.

Jacob: And this is the important distinction for this where, you know, it’s kind of you know, offer personalization.

Jacob: You think about it as more similar to kind of advanced AB testing.

Jacob: And so in reality, like for most apps, there’s not a hundred price points that makes sense.

Jacob: Like from a just like, think about a common sense perspective.

Jacob: Okay, like your fitness app today is 69.99 per year.

Jacob: Well, you’re probably not going to create a, you know, a $200 price point, you’re not going to create a 9.99 annual price point.

Jacob: And so we’ve got a range of like, okay, you know, maybe we’ll have 69.99, you know, with a trial, without a trial, with a discount, and then we’ll have monthly offers, you know, with trials and without trials, and then we’ll have maybe longer trials and maybe a, you know, some weekly offers.

Jacob: And so with a few products, like maybe like five, 10 different products, you can get quite a large combination of different offers for these users.

Jacob: And so this is how we think about it.

Jacob: And this is what we’ve seen work that like and these small changes can be a pretty huge improvement for everybody.

Jacob: And so this is typically how it works.

Jacob: And this, you know, works same way it works across web, you know, Google, and so this is and this is also better for everybody too.

Jacob: Like you’re you have complete control over what products.

Jacob: Like we’re not dynamically generating anything.

Jacob: You know, the marketing manager, the growth manager, product manager ultimately has full control over what’s happening, over what the offers are, can change and remove things, can override things.

Jacob: We really want this kind of explainable AI like.

Shamanth: Yeah.

Shamanth: And I would also add even within Apple, you can’t do 6421, right?

Shamanth: Because they have a very discrete price points only.

Shamanth: Exactly. Yeah, that’s also another factor.

Shamanth: So what from what you’re saying, look, there’s a very limited number of price points, but combined with the other factors that make up the offer, that can add up and those are that’s the universe of things that you are really picking from.

Shamanth: And with all of this, how do you think about fairness to users?

Shamanth: And I so also ask this because I remember

Shamanth: At least when once when I was in house, we just had a couple of customer complaints because basically saying, well, look, I’m just having to pay higher and my buddy’s actually paying lower.

Shamanth: And even though it was a small like one or two dudes, the team was kind of freaked out.

Shamanth: It’s like, are we setting a bad precedent?

Shamanth: Is this going to be a PR nightmare?

Shamanth: So, how do you think about that?

Jacob: It’s an important question and we think a lot about this as a team of, you know, what’s really happening here?

Jacob: At the core, typically what’s happening is that

Jacob: We’re typically maintaining your existing converters,

Jacob: Maybe converting a small percentage that have really high intent on higher value offers.

Jacob: And we want to get away from this like price is often a smaller lever than offer.

Jacob: And so simply showing one high intent user, instead of a paywall with a monthly and an annual plan, we put the monthly plan behind a see all plans page and show that higher intent user the annual up front.

Jacob: Or maybe there’s no trial on the annual.

Jacob: And it can be the same price across.

Jacob: And this can be a huge win.

Jacob: And so and then on the flip side, if you think about like on average, what percentage

Jacob: Percentage of new users actually convert for an app, like 5%.

Jacob: And so there’s a huge the opportunity to actually convert all those people that are churning and leaving and that you’ve paid for probably, right?

Jacob: If with kind of paid acquisition. These aren’t free people. I mean, if you have a large organic base, maybe, but most of the time, like if we can get, if we can go improve that kind of conversion rate number from five to 7% or five to 8%, like this is huge.

Jacob: This completely changes your unit economics, changes everything you’re doing, changes your ability to scale.

Jacob: And so a lot of times actually, that’s a pretty big opportunity as well.

Jacob: And so in one sense, like you’re kind of increasing access to your product and allowing more people to use it.

Jacob: You have to look, you’re building a business, building things isn’t free. You have to charge people.

Jacob: But most founders, most teams, if you told me, hey, I could have more people using my product.

Jacob: Like, you’d say, okay, yeah, that’d be great. I definitely want this.

Jacob: And so this is really what we see happening more. Than simply like there’s no price gouging, there’s no predatory pricing.

Jacob: It’s figuring out the offer that aligns with people because some people need a trial, they want to they need to test things out longer.

Jacob: Some people are ready straight for annual plans.

Jacob: Some people we can tell they’re from their data that they have a short-term use case, they’re just going to have to buy a weekly plan and that’s the only thing they’re ever going to convert on.

Jacob: And so really it’s more about aligning the offer with the user than anything else.

Shamanth: Yeah, and I think what I find interesting is and you’ve mentioned this a couple of times today, that the most important thing is the package and not the price in isolation.

Shamanth: And I think that is very, very under appreciated by myself too.

Shamanth: You know, because, yeah, because it’s easy to sort of think, oh, let’s just price test the hell out of this versus even the framing, the anchoring, the discounts, the trials, or the messaging for that matter, right?

Shamanth: So what you’re saying, like are you saying, you know, sleep better or wake up excited?

Shamanth: And I just made that up on the fly, right?

Shamanth: And so what you’re saying, a lot of these matter a lot more than the number that everybody sees on the screen.

Shamanth: Interesting. Interesting.

Shamanth: And how does this approach impact media buying and user acquisition?

Shamanth: Because again, the user acquisition manager is basically still continuing to see the first transaction primarily.

Shamanth: Or the trial, they’ve never obviously been able to see what happens downstream.

Shamanth: But now if you’re like, hey, some users have trials, some users don’t, some users have a higher price, some users have lower.

Shamanth: How does this change things for the media buyer if at all it does?

Jacob: Yeah.

Jacob: We typically see it doesn’t change things too much.

Jacob: When you look at the more sophisticated teams that if you’re doing a few million dollars a year, you typically have some level of sophistication and advancement and you’re probably not having one single product everywhere.

Jacob: You know, simply on the paywall, you probably have multiple options.

Jacob: And so there already people are buying different things, right?

Jacob: And so these are different price points, different plans.

Jacob: Then you probably have some discount winback offer that converts some percentage of people at a different price.

Jacob: And so if you think about right there, like this is what most teams deal with today.

Jacob: I think they don’t always appreciate that’s what’s happening.

Jacob: And that these different value signals are already being sent and you’re already dealing with this.

Jacob: And you try to deal with the best as possible.

Jacob: Now, obviously, if you’re sending trial start events as your, you know, optimization signal and we’re changing the number of products that have trials, right?

Jacob: This makes a difference if there’s fewer trials, but ultimately you’re in control of this and of all the possible offers.

Jacob: And so we also want to like I see more and more teams moving away from you know, a simple trial start event to kind of more predictive signals.

Jacob: And so this is kind of what our models are doing as well.

Jacob: We are optimizing towards RPU and LTV.

Jacob: And so this should ultimately help your ad buying long-term.

Jacob: In two possible ways.

Jacob: One, either your conversion rates overall are staying the same and we’re increasing your RPU, or we’re increasing both your RPU and conversion rates, which ultimately means more signals, more postback events to the ad networks of higher quality.

Jacob: And so generally like most teams, they benefit more from higher number of quantity of signals.

Jacob: Like typically just the more signals we can send back, this allows meta or whoever to find more users like this.

Jacob: Until you get to like pretty large scale, then that’s only kind of, you know, you’re doing a few million dollars a month maybe, then you can start pruning the events you’re sending back and send back higher quality events.

Jacob: You know, you can do some things like you know, the qualified trials a lot of people are doing now where you remove early cancellors.

Jacob: And so that’s some a bit of kind of refining the quality of signal you pass back.

Jacob: But ultimately, teams are already having mixed signals that are coming back to the ad network for different prices, for different products, for different trial lengths.

Jacob: And then if we can increase the conversion rates and the value of these, it should be neutral or net positive and this is what we’ve seen.

Jacob: We haven’t ever seen any real kind of negative impact.

Jacob: You know, we do for some teams we’ll start with like a roll out to a percentage of users.

Jacob: So you go, okay, let’s launch with 10%, measure this accurately, and then kind of scale from there.

Shamanth: Yeah.

Shamanth: And you know, what I’ve seen even with some of the larger teams that we’ve worked with is they operate and do a CPA target.

Shamanth: And as and when their LTV shifts, you know, so to that extent, then they don’t worry too much about what the revenue is as long as they have a LTV model to support the CPA target they’re hitting.

Shamanth: So to some extent, at least based on my experience, having a lot of price testing or price personalization isn’t super disruptive unless of course your conversion just completely crashes or just goes through the roof.

Jacob: Yeah, this is what we’ve seen as well.

Jacob: And we’ve had people go, oh, I price tested and my ads weren’t profitable anymore.

Jacob: It’s like, yeah, well, that’s because you were showing, you know, like 50% of your users a bad price that just people weren’t converting on.

Jacob: And so it’s like there’s this people get caught up in their heads like, oh, price testing affects my ads.

Jacob: Like, well, yeah, because if you show bad offers and people aren’t converting anymore, well, yeah, that’s a negative thing.

Jacob: And so but this is kind of it’s stuck in this old way of thinking of like there’s one offer that’s going to win.

Jacob: And ultimately these models aren’t showing the wrong offer to the wrong person.

Jacob: And so, yeah.

Shamanth: Yeah.

Shamanth: But even in a world where one offer would win, you would have the exact same risk that you test two things and one crashes and burns, you know, and your user acquisition is just not working.

Shamanth: Right? Yeah.

Shamanth: And you know, we’ve had a couple of blind spots if you will where we just had the UA seemingly evaporate and it just turned out that the product team just forgot to tell us that they were running these tests.

Shamanth: You know, and that happens more often than I would like to admit.

Shamanth: But you know, yeah, it’s a real thing that you have to be aware of the dependency between UA and price testing and even onboarding testing and stuff.

Shamanth: Excellent, Jacob.

Shamanth: This has been incredibly instructive and much like I said I did with your blog, there’s certainly something I’ve taken notes around.

Shamanth: I think the big one that’s really struck stuck out to me was that price in isolation is really not as important as price in the scheme of the package, the offer.

Shamanth: And certainly there’s just a lot more gems that you shared that I’m excited to dive into.

Shamanth: With that, I think this is perhaps a good place for us to start to wrap up.

Shamanth: But before we do that, could you tell folks how they can find out more about you and everything you do?

Jacob: Yeah.

Jacob: Please go check out botsy.com.

Jacob: I’m sure we can leave a link in the show notes.

Jacob: We’ve got you know, we have our price power podcast as well.

Jacob: We aim to kind of bring out insights and have world-class growth leaders on the podcast.

Jacob: Yourself to kind of share all their wisdom and knowledge to help other you know, subscription consumer founders grow their revenue.

Jacob: I also have retention.blog, a newsletter that also shares great retention monetization insights. I aim for super tactical stuff.

Jacob: And, you know, I would say if you’re interested in kind of personalized offers and a personalized model to figure out the right offer to the right user, feel free to message me on LinkedIn if you want to chat and learn more or you can schedule a call and we’ll be happy to chat more.

Jacob: I would say that, you know, even if you’re even if you’re not ready for a personalized offer model, like you said, Shamanth, you know, start testing and start iterating on your packaging and your paywall design, how things are presented, and you’ll start to see the possibilities and you’ll start to see under the hood that you’re under an AB test, there’s many different segments that perform differently.

Jacob: And I think whether you use Botsi or not, this is the realization we want people to have that your user base is not one homogeneous group.

Jacob: There are many different personas and segments that perform differently.

Jacob: And so I would say, you know, go back and look at some AB tests and try to segment and see what you can learn from there.

Jacob: So, yeah, people can go check me out, but hopefully also you can kind of take some valuable insights away.

Shamanth: Wonderful. Wonderful. And we’ll link to your blog, your website, and your podcast just as well.

Shamanth: And for now, thank you so much for being on the show, Jacob.

Jacob: Thanks. Yeah, this is great.