[00:00:00] Ryan Seams: We didn't really have a heavy PLG motion. A lot of this was done via weird custom contracting. In the move back to events later on our PLG motion was very clear and our pricing was very transparent with that. And so it did kind of give the [inaudible] to the developers. Like if you want to track this, great, if you don't want to track it, that's fine too. But like what an event is and you can make that decision logically versus in 2014, developers had a lot of challenges.

[00:00:32] Andrew Michael: This is Churn.FM, the podcast for subscription economy pros. Each week we hear how the world's fastest growing companies are tackling churn and using retention to fuel their growth.

[00:00:48] VO: How do you build a habit-forming product? We crossed over that magic threshold to negative churn. If you need to invest in customer success, it always comes down to retention and engagement. Completely bootstrapped, profitable and growing.

[00:01:02] Andrew Michael: Strategies, tactics and ideas brought together to help your business thrive in the subscription economy. I'm your host, Andrew Michael, and here's today's episode.

[00:01:12] Andrew Michael: Hey, Ryan. Welcome to the show.

[00:01:14] Ryan Seams: Thanks, Andrew. Excited to be here today.

[00:01:16] Andrew Michael: It's great to have you. For the listeners, Ryan is the head of Customer Success at Assembly AI, a speech AI company focused on building the new state of the art AI models that transcribe and understand human speech. Prior to assembly AI, Ryan was the senior director of customer success at Mixpanel. Ryan grew with Mixpanel over nine years where he started out as a support engineer. So my first question for you today is what made you take the risk of leaving Deloitte to join an early stage startup in 2014?

[00:01:46] Ryan Seams: Yeah, it's a great question. I think it kind of goes back a little bit to my university days. So I studied engineering back in college. And as I was studying engineering, I was really interested in the intersection between the hard sciences and business. Ultimately, that kind of led me to Deloitte and I thought, "Hey, let's go work at a company where you can do consulting. You'll work with a lot of big names and you'll understand how they implement technology and what that looks like."

[00:02:14] Ryan Seams: Upon starting at Deloitte, the project that I was assigned to, ultimately, didn't really, I guess, meet my expectations in terms of the technology we were working with. It really wasn't as cutting edge as I expected. And so, it led me to really think a little bit more deeply about what I wanted and the experience I wanted in my career. And that led me to think, "Okay, great. I'm at Deloitte. I'm working on this technology that I'd consider maybe a little bit legacy."

[00:02:38] Ryan Seams: What I was doing day in and day out was actually supporting that technology's support ticket queue. I'm actually helping diagnose issues with it and then go, of course, provide either bug fixes or actually send it into the development team. Tickets, essentially for the next sprint to go and resolve issues with that technology.

[00:02:55] Ryan Seams: And I thought to myself, “Hey, what I'm doing day to day is pretty cool, but the technology isn't as exciting as I want it to be. Maybe I should go explore some of these tech startups that are out there where I could apply some of these support skills in a role in a high growth tech startup.” And that kind of led me to go look at the market and find what else was out there.

[00:03:13] Andrew Michael: Very nice. And obviously what was out there was Mixpanel for you. You spent like the best part of a decade there. Maybe for the next question is what kept you there so long as well? Because I think your tenure is quite rare when it comes to early stage startups and what was it about Mixpanel that sort of kept you there all those years?

[00:03:32] Ryan Seams: Yeah, I get this question a lot. And I think, you know, ultimately if you're not learning, right, you're going to end up leaving and going somewhere else and seeking new passions. Right? I just told you why I did that when I left Deloitte. And I think at Mixpanel, ultimately there are very different, almost like chapters or seasons of my career at Mixpanel.

[00:03:49] Ryan Seams: Like I said, I started as a support engineer. I did that for about a year and a half. I then moved to being a solutions architect, working with some of our larger customers on their implementations and configuring all of their data strategy. At that point, there was the opportunity to move to London actually and open a whole new office for Mixpanel. And so I ended up moving to London for two and a half years and getting the experience of building an office in London and Singapore and Paris and Barcelona, and of course, scaling and growing the team there.

[00:04:18] Ryan Seams: COVID hit, that was a pretty big curve ball in all of this. And at that point, I was actually given the opportunity to move back to the US and lead our new and existing business sales teams for our commercial segment. And for a long time, I worked with sales, but there's no better way to actually understand the mindset of sales than going and doing the job yourself. So I took that opportunity. And then after all that, there was an opportunity to basically come back and lead the team that I started on, which was the customer success support and sales engineering team.

[00:04:48] Andrew Michael: Nice. Yeah. I think it's also a rare ability as well for people to be able to scale through those different phases of growth. I think like a lot of times you see different stages, like a lot of churn happening because people may be better at the earlier stage or they're at a later stage. And like I've seen this, like the changing of the guards happens at some point in early stage startups. And the ones that like really are able to adapt and learn, I think can thrive in this environment and grow with them. And I think if you can find interesting paths, the way you did, I can see how it's always exciting and something new, but familiar at the same time.

[00:05:20] Ryan Seams: Yep.

[00:05:21] Andrew Michael: That's pretty cool. And yeah, so like today we had a whole bunch of topics we discussed before the show that we could touch on. And I think like there's a couple of areas I think today we'll talk a little bit about, but the first one I was really keen and interested on is that I, myself, was all previous Mixpanel customer many times over, fan of the product, not a fan of the pricing model.

[00:05:42] Andrew Michael: And obviously now like at assembly AI, you've shifted into business as purely usage based pricing. So I'm keen to dive into this a little bit as well. Maybe just to set the context a little bit, you can give us a little bit of an overview of Mixpanel, what it does, how they manage their pricing and packaging and then now Assembly AI and the same thing. And then in the context of your role, how this makes a difference?

[00:06:03] Ryan Seams: Yeah. Yeah. And I bet a lot of our listeners have probably been a Mixpanel customer at some point in time, given the long tenure of the company and the fact that it pioneered kind of the event-based analytics model. So I think Mixpanel, right, when I started there, it was like 50 people. When I left, it was like 450. You can imagine during that journey, we had a lot of different iterations of pricing and packaging, ultimately, that came out to our customers.

[00:06:28] Ryan Seams: But at the end of the day, right, what we were ultimately charging on was either events, so essentially like, number of API calls, right, that we were ingesting from the end customer or what we called monthly tracked users, which essentially was like a proxy of events based on the total audience that you might be sending to Mixpanel. So if your website had 100,000 unique visitors, you would essentially have 100,000 monthly tracked users. And that, of course, correlated to some number of events.

[00:06:55] Ryan Seams: Both of those things, if you think about what those are, whether it's events, which is essentially API calls or these number of users, is kind of a proxy for the API calls, are ultimately based on the usage of what you're going to send Mixpanel. They're not based on, right, the number of seats or people that are going to end up accessing the platform later on. And so from a usage-based billing perspective, Mixpanel will have different like levers and packaging that goes along with that. But ultimately, we're charging you for the volume of data that you're sending to us.

[00:07:23] Ryan Seams: I think what's interesting from like a customer success perspective, right, is a business like Mixpanel actually kind of has like two axes of ROI. One is that usage-based model where you're sending data in. Can you ingest it? Is it available? Can I have access to it? The other piece though is more of the traditional like seats-based model where are people able to like build reports in the UI and are those reports accurate? And are they getting value out of the reports that they're generating, for example?

[00:07:53] Ryan Seams: And so while Mixpanel's pricing model was usage-based, a lot of times the ROI of Mixpanel was not just the usage-based pricing. And so from a churn perspective, right, you can imagine that can cause some complications there.

[00:08:07] Andrew Michael: Yeah, no, for sure. And I think like, yeah, to that point as well as like, you can be ingesting all the data in the world, but if nobody's looking or if nobody's using it, it's like, it's a waste of time. And I think there's obviously like complexity as well with analytics because they often like start out with the best intentions and then some point they always end up a big mess because there isn't like careful thoughts at the start to get things right. So you also almost have like this double edged sword as well like fighting against you. Like the more people use them, the less they like have these good controls in place. And then the more messy they become.

[00:08:38] Ryan Seams: That's something we used to talk about a lot internally, actually. Right? It's like the more people that are using analytics actually like the worse and the harder it becomes because you just have so many more events definitions that you can quickly create a mess at that scale.

[00:08:52] Andrew Michael: Yeah. Absolutely. And so the models as well, like I'm keen to understand as well from Mixpanel's perspective, because I know they changed over the years. And obviously like the first one, and I think just thinking through like from a retention perspective on a pricing and package level is that, I think when you first started out, was - and I'll say you now because you were there for long enough to be considered you after 10 years. But when you first started out, Mixpanel was really like events driven, if I remember correctly, the pricing and packaging.

[00:09:19] Andrew Michael: And I think the shortfall on that from my perspective was like, always felt like you needed to restrict which events you wanted to send to try and like avoid huge bills, which meant that like I wasn't like tracking everything I wanted to track. And then ultimately like not having the data available when I wanted it for the reports to generate. So it felt always like there was a gap in it there because "Okay, like I couldn't do what I wanted to do."

[00:09:42] Andrew Michael: Then on the other side, like things switched eventually I was like, "Okay, this is cool." Pricing and packaging now is by user, like, and send as many events about these users at once, but like, I'm only getting charged for the people I'm tracking. And that felt to me like, "Okay, this is what I want to do." Basically I want to understand what these people are doing and like, just let me understand as a paying per user. But then Mixpanel flip back again, I think if I'm correct. They went back to event base.

[00:10:06] Andrew Michael: So I'm keen to understand obviously like what happened during those transitions and like, what was the logic and thinking behind these changes? Because I'm sure like, obviously what I'm saying now is probably seem very obvious and those discussions and debates were had internally. So can you tell us how like what goes through the process of like making these decisions and these changes at the scale of Mixpanel?

[00:10:27] Ryan Seams: Yeah. Yeah. So to rewind at the beginning, right? We started with events. That's kind of what, you know, again, we pioneered the space. It's basically, for people who are listening that maybe aren't familiar, an event is essentially an API call that says, "What did this user do on some digital property." That could be viewing a webpage, opening an app, clicking a button, whatever that happens to be. And then of course, there's a bunch of metadata associated with that in Mixpanel speak that's called properties, but the browser you're on, the phone you're on, et cetera.

[00:10:54] Ryan Seams: When we started with events, that was like the most tightly correlated thing with ultimately the actual cost that we had, right, internally. And so we were trying to very quickly bill our customers based on the thing that was gonna ultimately drive our margins as a business. What we found though, right, if you rewind to like, you know, 2014 is product analytics was new. People weren't sure what an event was.

[00:11:19] Ryan Seams: Half the calls, we were like explaining events for 15 minutes just to like help the customer understand what we were talking about. Right? These are all people that are coming from basically like Adobe Omniture, Google analytics, right? They're used to like sessions and completely other dimensions for measuring some of these things. And so at that time, events were really hard.

[00:11:39] Ryan Seams: The other part from a success and support perspective, is you were kind of alluding to this like a case where you want to track more, but you don't want to get penalized for it. There's certainly that, but what we saw too, right, is like people were just making a lot of mistakes. And so you would just start some tracking and some event was sending rogue and all of a sudden you'd have, whatever, a million events overnight. And you've had racked up like a $10,000 bill and it causes this huge problem with the customer, right?

[00:12:06] Ryan Seams: Because they're getting this bill. We have to go, you know, to our support team. We have to issue a credit. And so there was just this tension between us and the customers constantly around like, "If I make a mistake, I'm gonna get billed. If I attract something that I'm not really interested in in the longterm, I'm gonna get billed." How do I balance as a customer, right, this weird equation where it's like, "I wanna track stuff and I wanna get value out of this, but I also don't wanna pay too much."

[00:12:32] Ryan Seams: And so the move to monthly track users was really intended to take that burden off the customer, right? And instead, put it on Mixpanel side to manage. And ultimately monthly tracked users, like I said earlier, right, was a proxy for the number of events that you're sending. It was just done in a way that hopefully was easier to understand to the customer.

[00:12:51] Ryan Seams: Now everything starts with the best intentions. You start off with that and it sounds really great. And ultimately what ended up happening though, right, is when you say monthly tracked user, the customer also has their own measurement of monthly active users. And what would happen in these cases, right, is the customer's monthly active user is defined by their data team and they have a very specific definition and they track in a very certain way. Mixpanel would then have their monthly tracked user number.

[00:13:18] Ryan Seams: And you also ended up with the same tension each month, but instead it was debating like, "Why do you have more monthly tracked users than I have monthly active users? Like is my tracking broken? How do we calculate this differently?" And so there was this constant push and pull of this discrepancy between the two numbers.

[00:13:36] Ryan Seams: Now, was it better than events? I mean, arguably, yes, because, right, you could at least track whatever you wanted and we could figure out this monthly tracked users thing later. But over time, what we saw, right, is it just became so complicated, especially as customers had a website, a mobile app on iOS and Android and all these different properties and their calculation for monthly active versus monthly tracked just didn't line up.

[00:14:00] Ryan Seams: Of course, this was like, I think 2017 or 2018 that we went and launched monthly tracked users. During that whole process, there was a whole crop of other product analytics tools that came to be. Events became very commonplace. Most people actually would be able to define what an event is today, right, versus 10 years ago, like I said, we were explaining this on calls. And so we ended up moving back knowing that ultimately, right, if you're a developer, and this maybe goes more to the usage-based conversation that we could have around AI tools as well, right?

[00:14:29] Ryan Seams: If you're a developer and you're sending API calls to a usage-based business, we should let you have the controls of what you're going to send and know very clearly what you're going to be billed for. And you can decide, right, is that thing worth sending or is it not sending? But I think that transparency part is really key in the early days.

[00:14:47] Ryan Seams: We didn't really have a heavy PLG motion. A lot of this was done via weird custom contracting. In the move back to events later on, you know, our PLG motion was very clear and our pricing was very transparent with that. And so it did kind of give the [inaudible] to the developer like if you want to track this great, if you don't want to track it, that's fine too. But like, you know what an event is and you can make that decision logically versus in 2014, like developers had a lot of challenges doing that.

[00:15:15] Andrew Michael: Yeah. Interesting. I can definitely see sort of how that would have caused friction and having this debate and discussion continuously around our numbers don't align because like, it's almost like an endless discussion internally at startups when they're looking at the different data sources and it's almost always like a waste of time, these conversations, but people always want to have them just because it's naturally always never going to be perfect just because of the different mechanisms. And there's like always a hundred different reasons why things could be the way they are as well.

[00:15:43] Ryan Seams: Yeah.

[00:15:44] Andrew Michael: So yeah, I can see.

[00:15:45] Ryan Seams: Yeah. It's funny you mentioned, I mean, that was one of our key churn indicators actually at Mixpanel, not the MTU versus MAU. But ultimately, right, if the customer was writing in being like, "This doesn't match my BI tool and like what we're displaying to our team." That was a pretty big red flag, right, that Mixpanel wasn't actually on some source of truth dataset and something that people were using to make decisions, right?

[00:16:07] Andrew Michael: Yep. That makes a lot of sense. You mentioned obviously then that this is a segue then into sort of like the usage based model. So Mixpanel then, just for clarity as well, like this was usage based pricing, but I think also what happened is like Mixpanel, it feels like it's a layer of complexity in it, but I also like it as like a layer of simplicity. And I think like, I wonder if a lot more AI native startups will be adopting something similar to Mixpanel because I think what Mixpanel just like, for [inaudible] and my understanding is Mixpanel had like sort of these tiers that you would fit in and you would have like an X number of events per month or per year.

[00:16:40] Andrew Michael: And that was your plan and you knew it was a fixed cost. And then if you went over that plan, there was overages, but at least you sort of always knew and you could choose if you wanted to do overages or not. I think if that came in at some point, but it feels like as well, like now with a lot of these usage based tools is that, at least my concern, like coming to picking a provider or vendor is like, "Okay, like, well, what happens in this scenario when we go over like our budgets and like, how do we set budgets on this? How do we know what it's going to cost us?"

[00:17:08] Andrew Michael: And like, I've, literally been in that scenario you described where we had an additional 10,000 bill overnight for Mixpanel because we sent the wrong events and it was firing continuously. And so, I think this is one of the things I'm keen to understand, like how you seeing this now in the markets like Assembly AI, fully usage-based pricing. What are some of those discussions that you're having with your customers now and how are they accepting this new way of pricing across the board?

[00:17:35] Ryan Seams: Yeah, yeah. I think maybe to touch on the overages part first and foremost, right? Overages are never a good thing for your customer. There is nothing involved with an overage that will ever be a positive for your customer. Overages are meant for your own internal team to have something as leverage to have a conversation with the customer around some sort of new agreement and type of agreement that they want.

[00:17:59] Ryan Seams: Personally, I think that way of working with customers is going away. And I think Mixpanel as well as Assembly AI have adopted a little bit different of a strategy around this, which is, "Hey, if you agree to," whatever, let's just make the math simple. "If you agree to 10,000 API calls for $10,000, right? Your price is $1 per API call. If you use more than that, we're just gonna keep billing you that rate a la carte. And, if you wanna commit to a higher number of API calls and get a cheaper price, please go ahead."

[00:18:36] Ryan Seams: But there's not this case where it's like, "We're now gonna charge you $2 for every API call because you signed some contract and now we have to have the salesperson have a conversation with you to set up some new agreement." And so I do think, especially in the age where developers are the ones buying these tools. They're the ones making these decisions again, putting the power into their hands. If you're the one that's charging them $2 just because they went over some arbitrary limit you set on your side, they're just going to go look at some other tool that's probably going to be half the price of that at that point in time.

[00:19:06] Ryan Seams: And so from that perspective, right, I think, if you can be transparent, like what is the rate they're going to pay for whatever this usage based thing is, what's going to happen if they spend more than that? And then of course, giving them plan types and packaging options of what does that look like if I want to commit to a higher volume? Can I get some discounts that scale with my usage?

[00:19:28] Andrew Michael: Yeah, that makes a lot of sense. I think to your point around overages, literally went through the same situation recently. Got an email from sales saying, "Hey, you've gone over an additional X amount per month." I was like, "Okay, well, let me go see what else is out there. Who's going to charge me half the price and make a switch." And yeah, I think it's not like a great position to be in where it's like, you're getting penalized for your success and giving a company more money. And when they've already sort of given you an agreed level.

[00:19:55] Andrew Michael: And to your point, I think the new models should be just say, "Yes, okay. Like this is the rates. Like if you go over it, you continue to pay this rate. You're not going to be penalized for giving us more of your business." So yeah. Interesting as well to see like Mixpanel backtracking on that and making changes there. Because I think they were the ones that sort of pioneered it as well, if I remember correctly at some point as well.

[00:20:17] Ryan Seams: Yeah.

[00:20:18] Andrew Michael: Yeah. So how are you seeing though, like this adoption then in the market? Like obviously like at Mixpanel, you saw different levels of friction and different problems with the pricing models. Like what are some of the conversations you're having now with your customers where there are flaws or holes in the pricing and packaging?

[00:20:35] Ryan Seams: Yeah. Yeah. So I think first and foremost, right, if you're working with developers on implementing any sort of AI tool, right? Their forecasting is just going to be all over the place. This is basically what it was like at Mixpanel 10 years ago when you're like, "What's an event? I have no idea." So if you're a developer, right, and you're implementing some, I don't know, LLM based feature and they're charging you per tokens, you're like, "I don't know. I have no idea how many tokens I'm going to have. I guess- [overlap]

[00:20:58] Andrew Michael: What's a token?

[00:20:58] Ryan Seams: I'll figure it out over time. What's a token? With us, at Assembly, we do simplify it a little bit. So instead of packaging it as something like tokens, we are charging like per hour of audio. So it's kind of more like the proxy type metric like an MTU [overlap] was to an event. But ultimately, right, even with that, you know, if you're a new developer launching a new product for the first time and bringing it to market, how do you know if you're going to have 10,000 hours a month or a hundred thousand hours a month? You really don't today. Right?

[00:21:26] Ryan Seams: And so I think that's the first thing, right? It's hard to forecast. So can you develop, you know, a pricing and contracting model that fits into that type of workflow? I think the second thing with that is, you know, making big, huge monetary commitments right now when, you know, frankly, like a quarter from now, every single person in your space will have launched a new model with new capabilities, with new things that you need to go evaluate is really challenging.

[00:21:56] Ryan Seams: And so if you go to someone, you're like, "Hey, let's make a $500,000 commitment." Like the first thing that person's probably going to do is go look at all of your competitors and be like, "I wonder if they won't force me into a commitment," or "Do they have better models," or like, "If I'm going to commit to this, I'm gonna have to go run a test versus all the rest of the market to go and make that commitment." And so I think those two things around like forecasting and commitment are true regardless of the customer you're talking about because the space is changing so fast.

[00:22:25] Ryan Seams: The pricing models are hard to understand and nobody really knows like, you know, with any sort of certainty, strong answers to where they're going to be, you know, a year from now in terms of what they want to commit and the pricing that they're ultimately going have with them.

[00:22:37] Andrew Michael: Yeah. And I think even more so in your business, like as you said, like we're literally like almost every day, not even every week, like new models are being rolled out and updates. And so I think from like a customer perspective and engineering, like you don't really want to make these long-term commitments because you want to always like be on the like the frontline and like utilizing the best service that may not necessarily be like the product that you're committing to for a year and see sort of how that being like a new reason for churn.

[00:23:05] Andrew Michael: I think this is also like, there's a lot of question around like a lot of these AI startups seeing hyper growth, like insane numbers never been seen before. And I still see a lot of questions are "I can't wait to see the churn numbers," and like this debate. I think from my perspective, it's probably like way too soon to even be like talking about these things. And I think naturally like it feels that there will be a high churn rate. But I also think like these types of business, other things come into consideration like the reactivation rates. You may see huge churn, then a week later, people reactivate. A month later, people reactivating.

[00:23:40] Andrew Michael: There's a number of different things. I think also because of this insane growth that they are seeing, churn just gets masked completely. There's almost like - it's very difficult to understand what churn is really looking like just with this growth that they do see today. So yeah, I think it will be interesting. What's your perspective, being inside one of these companies, seeing the numbers, seeing how Mixpanel was like more of a B2B traditional SaaS versus now like an AI first company providing services in this constantly evolving environments?

[00:24:09] Ryan Seams: Yeah. I mean, if you were just to look at the numbers with no filtering applied and you came from like a B2B SaaS company, you'd be like, "I don't understand what's happening here. Like these numbers are scaring me." And the reason I say that right is like, keep in mind, there could be people who are like out of random Gmail, they want to like, you know, pay 50 bucks. They just want to try your product and see if it's a good fit for them. And maybe it's not and they leave or they used it for some hackathon project and they're a student and they're not back the next month.

[00:24:35] Ryan Seams: And so if you look just in aggregate at the numbers, right, things like logo retention are going to be way lower than you would ever expect from a B2B SaaS product, right? That being said, you have to be smart about how you calculate these rates and how you actually show them to your investors and how you think about them internally when you're showing them at all hands and company presentations. And so you have to start to apply filtering.

[00:24:59] Ryan Seams: For example, what's a use case of this? Well, someone on a Gmail account is probably very different than someone on a corporate account. You should probably look at those two cohorts differently. Again, another layer of filtering. We should probably filter out people who literally just tested and never did anything else versus people who actually deployed the software somewhere and see what their churn rates are. And so, you know, we have our own ways internally of calculating, right? Was this a test and an evaluation or is this person in production? That's another way that you might segment that.

[00:25:27] Ryan Seams: And so ultimately, if you think about like your users experience or customer experience and you start to layer in that segmentation, you actually get to metrics that are very similar and look a lot more like what you think about from like a B2B SaaS traditional lens in terms of logo retention, gross retention, net retention, et cetera. But you do have to do some layers and level of filtering based on what your customers are actually doing with your product before you start to see those metrics actually mean something and you can trend and track them over time.

[00:25:58] Andrew Michael: Yeah. And so like, I think what you're saying as well is like the use cases and the different types of users are like exponentially greater than like, let's say like a B2B SaaS product for analytics tracking where it's like in this new AI paradigm, like it's almost anyone and everyone like trying to give it a go and not anyone and everyone's a good fit customer. And you really need to understand who your ideal customer profile is, like who you're building for, and then understand what churn and retention looks like in that audience for you to understand, "Okay. Like is the product sticky or like, have we just got like a sinking ship that we're waiting to go down?"

[00:26:32] Ryan Seams: Yeah, and keep in mind there are a lot of these products are a lot easier to test and implement, right? And they usually don't have any upfront commitments. If I was going to test Mixpanel, that's a pretty big ask. I'm going to have to spend some time, send some data, do a bunch of work. With these, it's like, I mean, you could literally not even have a coding experience right now and use some of these tools. Like, "Cool, I have an Open AI API key. I went and ran this prompt and wow, I built a website all of a sudden, right?" And like, that's a user now for them, right?

[00:27:00] Andrew Michael: Now it has voice involved [inaudible] [overlap]

[00:27:03] Ryan Seams: Is that going to be a long-term user for them? Probably not. But, you know, if you're thinking about something like a logo retention metric, well, that person's probably going to churn next month and you don't want to be looking at them as potential high value users for yourself.

[00:27:16] Andrew Michael: Yeah. I think it's also like the market just expanded almost exponentially over the last like 12 to 18 months of like, who would be accessing the software now and who would be doing it. And like, as you say, as well, like I think a lot of them will come back and also like maybe like I did the first time around was a nice, fun weekend project. And then like two, three months later, I say to a couple of friends, "Hey, look. You have engineering experience, let's do something together." And then, I use Assembly AI before I like it to come back.

[00:27:42] Andrew Michael: And so, yeah, I think like there's definitely going to be like this in this new wave like my assumption would be that reactivation is probably going to be off the charts for most of these businesses, like compared to traditional SaaS businesses because of this like periodic use case testing things out trying. When something hits and sticks, they become sticky, but until then it's just like experimentation.

[00:28:02] Ryan Seams: Yeah. I mean, you can see in our own charts, right? Whenever we have our own launch, you can see reactivations go up. When your competitors have a launch, your reactivations go up because people are paying attention. They're all in the same spaces. They're seeing what everyone's releasing and everyone wants the best model for their use case, right? Ultimately, like if I have a specific use case, I want to pick the best model out there that works for that. But it is really hard to go and evaluate that. so people are just like constantly running additional evaluations over and over again, right, to really make sure they do have the right fit for whatever you're trying to achieve.

[00:28:36] Andrew Michael: Yep. And you mentioned something as well at the start, it's like the audiences you've typically worked with has been EPG so engineering product - EPD, and design. How do you see these audiences differing now versus let's say like 18 to 24 months ago? Like have you noticed any changes in their buying behaviors of software and like the discussions you're having from a customer success perspective of like, let's say like pre AI boom era versus today? Is there any significant changes you've seen selling or working with these audiences?

[00:29:07] Ryan Seams: I mean, I think when maybe they first started buying AI tools, they expected it to be like B2B SaaS. And maybe that's how the vendors even treated them at the beginning, or at least some of them. Right? And I think over time, what they've realized is like the way that you account for these tools is different. The way that you should be buying these tools is different. The types of agreements you should be signing with them is different. And so ultimately now, right, I think people get it. It's usage-based, based on whatever you're going to spend. And you ultimately just want to get like the best rate based on the amount of usage that you're going to go and send to that vendor. And literally with like the least frills attached beyond that.

[00:29:48] Ryan Seams: And so I think what we see with our customers, right, is like - when you're usage-based as well, you have to keep in mind, a lot of these companies are building a product with a bunch of AI models under the hood. All of those models are going into their own costs for their own product, and they're trying to make a margin-positive business at the end of the day. You're like probably one piece of that puzzle. And so what they're really trying to understand is like, "If my usage doubles or triples, what kind of cost profile can I get that it still makes my margins positive and looks good to my own end investors at the end of the day?

[00:30:21] Ryan Seams: And so those are the types of conversations you're having. I do think there's still a lot of similarities though, right? Like EPD doesn't like to be sold to. They don't like to do a bunch of negotiation. They don't like to do a bunch of frills and calls and meetings. Like they just want to understand very transparently, like "What's the pricing model? What happens if I do what I told you I'm going to do with my own usage and like give me the best price for that without a bunch of like, you know, back and forth and bells and whistles attached to that."

[00:30:49] Andrew Michael: Yeah. Like Elon, like, let's move to the end. No negotiations. Nice. [overlap]

[00:30:57] Ryan Seams: Here's an example that's helpful. Like we work with a lot of our customers in Slack. EPD, that's pretty common, right? They want to be Slack. And, you know, we often ask them, like, "Hey, you know, do you want to go to call and talk about this? Or do you want us to send some options in this Slack thread? And like, you can decide." And it's almost always like, "Yeah, just send me some options. I'll review them and get back to you with questions." Right? It's not a lengthy process.

[00:31:18] Andrew Michael: Process. Nice. So I see we're running up on time. I want to make sure I ask a couple of questions I ask every guest that joins the show. First question is like, what's one thing that you know today about churn and retention that you wish you knew when you got started with your career?

[00:31:32] Ryan Seams: There's so many things, but I do think one of the very simple things you can do, right, is pattern matching. So trying to understand what's happening within different ways that you're engaging and looking at your customer base and taking whatever, like the things that are working and just like scaling those. And oftentimes those things can be very simple.

[00:31:50] Ryan Seams: And so for me, for example, we just talked about EPD being in Slack. Adding an EPD person to a Slack channel is probably the best thing you could do to improve your retention. And it sounds so simple, but just find what are those little patterns and things that people are really enjoying about your experience and, you know, overindex on those.

[00:32:08] Andrew Michael: All right. Very nice. And next thing is like, we'll take it from the customer success lens, I guess, but what's one thing that you wish more people would ask you about customer success, but they don't?

[00:32:22] Ryan Seams: I think if you go look at customer success, there's a wide variety of use cases out there. You could be very commercially driven, almost like on the sales side. You could be more technically driven, which is usually the roles that I'm working with where you're working on implementations and code. I think specifically within that, I wish more people would just talk about how like the engagement strategies and the jobs and the day-to-day are just completely different for those roles.

[00:32:47] Ryan Seams: And, you know, for some people, QBRs are great. For my audience, QBRs are horrible and the role is the same. And so it's really important to have that lens when you talk about customer success. There's a huge variety of roles really under the surface there and the tactics and the things that you do day-to-day are very different amongst those.

[00:33:04] Andrew Michael: Very interesting. Yeah. I think it's something not often spoken about, but I think it's also because customer success is relatively in its infancy compared to all the other roles that exist today and a lot more well-defined and discussed and debated. And it's great that people are sharing these things.

[00:33:22] Andrew Michael: So Ryan, it's been an absolute pleasure today having you on the show. also want to thank Advith Chelikani from Pylon, who was a previous guest as well on the show for the recommendation. It was a great conversation. Is there any sort of final thoughts you want to leave the listeners with before we wrap up today?

[00:33:37] Ryan Seams: You know, more than anything, if there's anything I could do to help, if you have feedback, if you want to engage, find me on LinkedIn and more than happy to have a follow-up chat. But yeah, I mean, honestly, if I had any parting piece of advice, treat your customers like you want to be treated, put yourself in their shoes, think about the tactics and the things that you're going to them with. And ultimately you should really find a great customer experience based off of that.

[00:34:01] Andrew Michael: Very nice. A nice way to end the show as well. And again, thank you so much for joining us. For the listeners, we'll make sure to leave everything we discussed today in the show notes so you can check that out there. Ryan, I wish you best of luck going forward. Thanks again for joining.

[00:34:13] Ryan Seams: Yeah, thank you. Appreciate it. It was fun.

[00:34:15] Andrew Michael: Cheers.

[00:34:23] Andrew Michael: And that's a wrap for the show today with me, Andrew Michael. I really hope you enjoyed it and you were able to pull out something valuable for your business. To keep up to date with Churn.FM and be notified about new episodes, blog posts and more, subscribe to our mailing list by visiting churn.fm.

[00:34:43] Andrew Michael: Also don't forget to subscribe to our show on iTunes, Google Play or wherever you listen to your podcasts. If you have any feedback, good or bad, I would love to hear from you. And you can provide your blunt, direct feedback by sending it to andrew@churn.fm. Lastly, but most importantly, if you enjoyed this episode, please share it and leave a review as it really helps get the word out and grow the community. Thanks again for listening. See you again next week.

‍