Andrew Michael
Hey, Steven, welcome to the show.

Stephen Levin
Hey, Andrew, how you doing?

Andrew Michael
I'm very good. Thanks. It's great to have you on the show today. Thanks a lot for agreeing. For the listeners out there. Stephen is coming from a very heavy data and analytics background. He was a previous VP of growth and analytics at speak. He's been head of analytics at Cisco Systems. He's a startup mentor. He has his own data, freelance data analyst and growth hacker think analytically where's the owner. And he's also now the special projects for the CEO of Zapier, which is also obviously had a big influence on a business intelligence team and data analytics there. Steven, I'm very intrigued. What is the special project for the CEO Zapier entail?

Stephen Levin
Sure. So for us at Zapier and my role in particular, I handle project sort of in two classifications. One is something that's important and under resourced at the company. So for a while, we were short few recruiters, so I was doing recruiting. And the other classification is things that are extremely broad, and would otherwise sort of have to sit with the CEO, but on a tactical day to day basis benefit from having someone other than the CEO actually pushing that forward. So good example, there is our company, okay, our process, strategic goal setting, and sort of goals around the org. Everything in company strategic goal setting ultimately rolls up to the CEO. But from a tactical perspective, it helps to have someone other than Wade pushing forward working with the team's iterating on versions, one, two, and three of their goals. So that way, it only has to sign off on the final, you know, before version. So those types of projects that are either extremely broad, and would roll up otherwise to the CEO, or that are just important and under resourced is what I end up working on.

Andrew Michael
And how did you end up making this transition? Because I think prior to that, obviously, you were working with marketing analytics with the data team, like what, when did you make the transition? And why?

Stephen Levin
Yeah, so this role came up, mostly because it was something that I was looking forward to trying to do. I'd seen other organisations that have something in this vein, special projects, potentially business operations, Chief of Staff, all of those types of roles could have a flavour like this. And so my initial introduction to wait actually was pitching him on this type of a role. And when we talked, he said, Well, it sounds like you're actually a marketing analytics person. And he was right. So he's like, all right, well, we're hiring and marketing analytics, why don't you start there, and we can continue to talk about this later. And then a year or two down the line from that, he started to see a potential benefit to having someone focused like this. And so that's how I ended up moving over.

Andrew Michael
Very cool. It definitely is an interesting role. And like you said, I've seen it position in different ways as well like, such as the Chief of Staff for operations. And it probably puts you in exciting position as well to be able to tackle multiple different problems at different points as well.

Stephen Levin
Yeah, and it's very exciting to be able to see you know, how you use data all around the organisation. So we have a pretty sophisticated data team at Zapier. But to just have another voice, sort of working one on one with a lot of different teams and advocating for them to use data in setting their strategic goals for the year, really is a cool position and voice to have any ordinary. Absolutely.

Andrew Michael
So today, I wanted us to focus specifically on metrics and analytics when it comes to churn and retention. And I want to just maybe get started off with, obviously, you've worked with lots of different companies when it comes to helping them with their metric stack and looking at the different key metrics that they should be focused on. When it comes to churn. What are the typical metrics you advise startups and companies to focus on? And how do they go about measuring the churn?

Stephen Levin
Sure, so churn is actually a fairly nuanced concept and metric that a lot of people try to capture all in one number. And that's not necessarily wise turn should be broken out into several different concepts. So the first version of that that you would think about is account churn or logo turn, typically, that would mean you have someone who's paying you, and then they stopped paying you. So that account that business turned. The second piece that we consider is user churn. So you can have accounts or companies that have multiple users. And so a, an account that has 30 users is substantially more of a hit to your business when it turns than an account that has one user potentially. So you can also look at user churn which is similar to account churn. Third, you want to look at revenue churn. And so that's the same idea. But instead of waiting every logo or every account the same, you actually calculate what percentage of your IRR your recurring revenue churned during a given period. And so revenue churn is potentially the most important one and the one that you that I recommend thinking about the hardest, because that really waits for your larger accounts that you sort of need to keep around in order to continue building you can definitely get a revenue turn number that helps you focus on those larger accounts.

Andrew Michael
So if revenue churn is the one that you agreed to, you recommend people focus on why would somebody want to track customer and logo churn to begin with them?

Stephen Levin
Sure, so they tell slightly different storeys. User churn account churn, oftentimes, you may even not look at their the fact of whether they are paying you or not, you might look at user churn or account churn on a usage basis, especially if you have a freemium product like Zapier. So we have a whole bunch of users that don't pay us anything. They're using our free plan, and we hope that they'll upgrade later. But when we look at user churn, we often look at activity churn instead of paid account turn. So if they stop using the product, that is another variation on turn, where you have to count users because you don't have any revenue associated with those folks.

Andrew Michael
And I think that's something a topic where I want to chat a little bit about later is measuring usage and when somebody becomes dormant versus churn, but on this topic, then so just continuing the discussion with the different ways we're looking at lowering customer churn, we're looking at revenue churn in revenue in terms of the way we look at revenue churn, how do you typically recommend companies go about setting this up? What are some of the ways that you would typically want to segment revenue churn by and how would some of the reporting go when you presenting to the team or to exec within the org?

Stephen Levin
Sure, so the first thing that you want to think about when you're talking about revenue turn in particular is one big variation. So you haven't gross churn, which is just your accounts that stopped paying or that downgraded. And so you take the dollar value associated with that downgrade or with that cancellation, then you divide by you're starting a IRR. So whatever period you're measuring over, you take the gross churn divided by the starting a IRR, and that's your gross churn number. NET churn takes into account your upgrades as well. And so for a lot of businesses, you have this upgrade path, people aren't just paying or not paying, they have a way to go from paying to paying a little bit to paying a lot. And one of the typical numbers that you'll hear in SAS conversation is your net churn number. And in particular, net churn can be confusing because it can actually go negative. So net negative churn is the concept where your upgrades of the accounts that stick around, actually outweigh the losses from downgrades and cancellations. And in that scenario, your company is going to grow even if you have no new customers coming in the door. And that's an extremely powerful position to be in. In the SAS world.

Andrew Michael
One of the most powerful it's probably the holy grail in the SAS world.

Stephen Levin
Certainly, as

Andrew Michael
we've talked about this in a couple of different episodes, I think one is with Jana pesto from prod pad on the episode we recorded in the actually had recently passed, and it negative churn. And actually in the last episode with David Scott, he talked about the power of net negative churn and as an investor why it was sort of that lightbulb moment when you realised how powerful SAS businesses was, was when he understood this concept of net negative churn. And even if your company doesn't add any new customers, it will still grow and continue to expand due to the retention and negative churn. The next thing is well then on this is when we look at sort of churn and retention. A lot of the times I think people tend to potentially get confused when it comes to the annual versus monthly churn rates. And maybe you want to talk us through the different reporting on that and the math behind it. And when would want to be looking at both types of churn? What are the benefits of?

Stephen Levin
Yeah, so one of the pieces of churn that I see get mistaken, just in casual conversation typically is the difference between monthly churn and annual churn. So if you're thinking about your business, you're probably thinking about monthly churn numbers, especially younger companies, how many people are leaving during a given month. But because the range of churn values is so extraordinarily high, and all the way from net negative churn to net or gross turns in the five or 10% per month numbers, you can actually get confused between monthly turn calculations and annual term calculations. And more than once I've gone into a business where they were trying to benchmark themselves and understand how are we doing relative to sort of all other SAS businesses and they say, Okay, well, we have 5% term, that's great. But in their minds, they were saying we have 500 monthly churn, and we think that's great. When in reality 5%, monthly churn means an annual churn of something on the order of 45, or 50%, which no longer feels so good. And so it's worth making sure when you're having these types of conversations with people in your business, or with your investors, or anything along those lines, that you are clear to say, a churn and associated time period, you could actually look at it daily or weekly, although that's pretty a typical, you, but you really want to make sure that you're clear when you're reporting these numbers out. Yeah.

Andrew Michael
So definitely, I think that's, it's very easy to confuse the terminology and understanding and definitely an early stage when you're just getting into SAS and looking at metrics, it's critical to understand the difference and the impact that the monthly versus yearly churn rates have, because it grows exponentially pretty fast, like you say, 5% could end up being between 40 to 50%. itself, or the yearly basis. So thinking about that as well, like yearly vers, monthly churn. And when we think about yearly versus monthly contracts, what is some of the things and the impacts that you see businesses have when it comes to the different contract length and periods that they do? So have you looked into the churn rates and the comparisons when it comes to yearly versus monthly, and how well companies fare when they actually have different plans if its yearly or monthly?

Stephen Levin
For sure. So there's almost always a difference in the behaviour of these types of contracts. And it's unclear and you sort of have to look for yourself as to whether it's a self selection bias, the people who are willing to commit to a two year contract are more likely to stick around and therefore their turn rate is better. But one of the things that makes you think that annual contracts are probably a net benefit to the business, that you will oftentimes see a large drop in retention or increase in churn exactly when your accounts are one year old, if you have one year contracts, and you'll actually see a notable Cliff in your data when you're looking at this. And so I really encourage people to take your turn numbers or the inverse your retention numbers, and look at them based on age of account. And so if you have a graph that has the age of account on the x axis, and then net retention on the y axis, you'll typically see a descending line China, people, you know, leave over time, broadly speaking. And you'll start to see a cliff where a bunch of people, we, if your contracts are a month, or a year or three years, you'll see clips at the end of those contracts as people roll over, that indicates that they would have turned some time earlier. But they just let their contract expire. So annual contracts are probably benefit to the business. However, the other thing that these that this type of chart really gets to show you is one if you break it up by monthly cohorts, so is my April cohort doing better or worse than my March cohort is a really interesting question over time, you can see at 30 days, what was the retention of the April cohort? And what was the retention of the march cohort, how are they doing? Hopefully, your cohorts are doing better. And it also lets you understand, if you get what I have heard called an Evernote smile, Evernote, maybe a little bit of a Data Reference in this moment, but for a long time, they would actually see a huge fall off and activity over the beginning of a user lifecycle. And then it would go flat for a long time. And basically, if you still stuck around, you would stick around forever. And then it would tail back up again, indicating that people would sort of try Evernote, not really get it, it would hear about it again. And then they would come back. And so their cohorts that we're three years old, actually had more users active in them than their cohorts that were two years old. And that's another sort of holy grail of fast if you can reactivate users that deep into their life cycle, you're really doing something right, in the long term. And so I've heard that called the Evernote smile. But anyone looking at a retention chart, you would love to see that your accounts that are really old are actually picking up steam again.

Andrew Michael
Yeah, and and obviously the other case, where you start to see that small is with the expansion revenue. So you may have less customers over time. But if those customers are spending more money, you tend to see that upward curve if you're looking at the med centre or right. Yes, and that is definitely interesting. I think it's something like we had hot show, actually, quite recently, maybe four or five months ago introduced to yearly plans. So I like that you mentioned sort of that self selection, like bias in the sense that people that typically are going to select a yearly plan are probably the ones that are going to be less likely to turn to begin with. But it's also interesting, looking at cohorts, as you mentioned, and really looking at sort of what does that drop off. So maybe let's dive a little bit deeper into codes. Because I think it's probably one of the like the best ways to actually measure and track churn itself and a lot of other metrics. Maybe you want to just talk us through a little bit more detail around the concept and why it's so critical to look at cohorts when you're looking at things like channel retention.

Stephen Levin
Sure. So there are two big pieces that I really like to look at here. One is the one that I alluded to earlier, he's if you show the retention of each cohort over time, it really gives you a sense of how are the newest customers that we are helping fairing compared to the existing customer base or compared to earlier cohorts. And if assuming that your product is improved, your onboarding experience is improving your copy is improving all of that material should mean that your newer cohorts performed better than your older cohorts. And if they're not, you really need to understand why. Now it's possible when you think about Crossing the Chasm that you're getting into the early majority instead of the early adopter, or something along those lines where your newer cohorts may not perform as well. But there are a lot more of them. And so it's a beneficial targeting strategy anyway. But if you see that happening, you need to be proactive and really understand, why are our new cohorts not doing as well Didn't we screw something up and just understand what's going on behind those numbers. The second piece is particularly to understand that your churn rate will almost always be much higher for brand new customers than it is for established customers. And so when you're thinking about it all in turn rate for your company, that can be informative, but it's extremely important actually look at the churn rate of your customers younger than say 30 or 90 days compared to your churn rate of established customers who are older than that cut off. And oftentimes, you'll see the the early customers with a churn rate that's literally multiples higher than the churn rate of the existing business. And that's just because when somebody signs up, they may not have gotten all the value that you were hoping that they would get. They were hoping they were they would get. And so the likelihood of churn is just much much higher when account is young and fresh than it is at any given point later in their user lifecycle. So if you break the churn calculation calculations, in no other ways, you should absolutely break it down into young customers and established customers because those churn rates will give you different information.

Andrew Michael
Absolutely. I think it's typically like from a lot of the the previous guests we've spoken to it says first 90 days are critical when it comes to turn. And you definitely see a big distinction between like the first 90 days churn rate for is after that our customers also found a very interesting, it's a super interesting point that I think is often overlooked that you mentioned is that understanding how churn can be impacted by the type of customer you're going after and the stage that you companies in so we talked about this in a previous episode as well with Julie from drift, where they show notable differences Well, in both drift and HubSpot. Once that starts to cross the chasm and move out from the early adopters to the early majority. Is that something that you saw at Zapier as well? And like, how did you know that this was it was happening? Was it a previous company where you came up with this insight?

Stephen Levin
Yeah, so I mean, we certainly see different classifications as much as possible when we can break out customers into these different personas, we do see differences in their rates. And so understanding and even trying to distinguish among your most recent cohorts, who are we talking we're getting from our sort of existing classic, best customers versus Who are we trying to stretch? And maybe you can distinguish that based on their usage patterns, maybe you can distinguish that based on the plans that they self SELECT INTO. But we do our best to understand the churn rates of those different user personas in specific.

Andrew Michael
And do you see big differences between your personas that you have,

Stephen Levin
when it comes we do? Yeah, and it's not specific to churn and retention, we see differences in those personas across most of our metrics in terms of onboarding and activation. Your you really are always trying to improve those numbers, generally speaking, but you may end up targeting a big group of people who don't perform as well or don't click as naturally with your marketing copy, for instance. And so you just have to get better at targeting the right copy to the right person to get them stick around.

Andrew Michael
The touches on the lives of previous topics that we've discussed as well, I think, envision, similarly as will had this realisation of the different personas and how critical it was to treating them differently when it came to the marketing messaging to the onboarding, to the adoption rates, because they saw vast differences between the types of use cases and personas. So the other thing I want to talk about, then you mentioned in terms of like looking at cohorts, and why it's critical, because you can really get that picture and feel of how your team is doing, or your changes impactful. I think one of the challenges when it comes to churn, like, as you alluded to, in the beginning, is it's such a nuanced problem that has impact. So like so many different aspects can impacted. So you could make an improvement in support, you could make a product change, you could introduce new pricing and packaging. And these changes are happening throughout the organisation all the time, within different teams, when it comes to tracking the performance and the improvements or like moving back over time, like, how do you go about attribution? And when you start to think about like the impact specific changes are making? Is there any sort of ways that you've looked at tracking this and have a good system in place that can sort of, at least to some degree attribute, the changes that are happening within the world to the change you see in turn and attention?

Stephen Levin
Yeah, it's a great question. Turn in particular is hard here. So I'll give you a different example. Zapier is quite good, actually, with our experimentation infrastructure. So if we make a change, we run an experiment. And then we can say, of all the people who saw either version a or version of this experiment, how did they perform on some next step of the user journey, the benefit to a framework like that is that you can see very quickly did this person succeed or not, in your particular conversion event, where turn is so difficult is that conversion event is a very long time away. And so you end up making a change and hoping that you increase, let's say, 90 day retention, well, you won't be able to measure that for 90 days after your experiments finishes. And that makes for a really long experimental cycles. And at a young company, it means that you're moving too slow. And so where the path that we have chosen recently, is to try and figure out what the leading indicators of turn are, whether that's a behaviour or a visit to a particular page, or a second or third or fourth use case. And then use that in lieu of experimenting directly on the turn rate, to say, did we trigger an event that is likely to increase their long term retention? And that's not quite as good as saying exactly increased their long term retention, but it's much faster, and it allows us to experiment much more quickly. That said, we do always go back and look 90 days, 180 days, 365 days later, to understand, did these cohorts actually retain differently?

Andrew Michael
Yeah, so and different things like turn itself is really not put metrics. It's made up of multiple different inputs, input metrics, when it comes to sort of those input metrics or themselves like, what process Did you take to begin with to understand what would those leavers be that you believed would have an impact on retention? Did you do any sort of internal analysis and understanding of what those metrics or input should be that you would want to improve and tracking? How did that go about internally?

Stephen Levin
Yeah, so that's a great question. This is sort of an exploratory research project. That we, our data team has an explicit mandate to look where no one else is looking. And so one of the things that the team tried to do was to say, hey, are there indicators, potentially as soon as their first 14 days that would say, yes, this user is likely to stick around for X number of days later. And we found several behaviour patterns that were correlated with longer lifespans. So in particular, there are a couple of key things that a person onboarding to Zapier can do that will indicate to us that they are likely to stick around for a long time. Now some of that are repeat users say you leave the company that you're currently at you come back and use that beer again, you're much more likely to trigger all of these metrics that we're talking about that say you're a long term user. But a lot of those are just about using the product effectively. And so we actually turned it into an onboarding project and said, Hey, if we can get people to use the product really effectively, they're more likely to stick around. Now, that doesn't seem mind blowing. But when you can actually start to cherry pick off the pieces of what using the product effectively during onboarding actually means, then you have an opportunity to increase retention by focusing upfront.

Andrew Michael
Yeah, and not needing to sort of building in you features, right, but it's really about focusing on getting them set up correctly. But specifically like those, what using the product correctly means I think like for different companies, this can as well get quite nuanced and trying to understand and like, if you think about the case of Zapier is fairly similar to I think hi char as well, where there's multiple use cases that people can come to use the tool for, there's multiple ways in which they could use the tool properly. If we were in inverted commas. How do you go about figuring out, like which others actions is key actions that you want to prioritise? So for your onboarding, like how do you know which other main actions that people need to be taken first? Like, I don't know, if you've looked into your case? Like, are there any specific steps that you find the more successful people set up? Like, have you gone to sort of granularity in that? And what was your process behind it?

Stephen Levin
Yeah, so my statistics friends on the data team will not love my answer here, but I'll give it to you realistically. So what we're really looking for is a correlation, not causation, it's impossible to say without an experiment, whether doing behaviour a actually caused you to be more likely to retain. But what our preliminary analysis showed is that there's a correlation between taking these behaviours, yes, specifics, app specific use cases, multiple use cases, all of those things are obviously beneficial. And then we would try and guide people to do those things. Now, we are making an assumption here. And it's important that you understand when you're doing that, but that we're saying, we think these behaviours are actually causing you to stick around longer. But that may not be true. And so you have to continue to watch this stuff over time, just because certain behaviour patterns are correlated with increased retention doesn't mean that they're causing it. And so when you actually take an action to encourage more people to do those things, you have to continue to monitor whether the retention rate increases in the way that you hope it does, or whether it doesn't.

Andrew Michael
Yeah, and when you're doing this correlation analysis, what is sort of the methodology like take us through one very specific case of how you identify correlation with a specific action that a user takes, like, what is the formula that you're looking at? Like? How are you measuring the strength of the correlation?

Stephen Levin
Sure. So for better or worse, there's actually opportunities in most small businesses where the impacts are so strong, that you don't need rigorous statistics. So when we were talking before about these retention charts, and why I'm such a big fan of them, this is another reason. So you can actually take the same retention chart cohort chart that we were talking about. And instead of dividing it by, say, the month of the cohort, you divide it by some other action. And so we did several of those things, whether they had multiple zaps, whether they had usage on multiple days, whether they had usage in certain categories, or use case buckets. And when you just break up that retention graph into segments, along any of these axes that we're talking about, you can actually see extremely easily that yes, this one particular bucket performs way better than anyone else. And so you don't need in that case to say like, what is the statistical strength, it's just visually apparent that the differences double or triple. And so when you're talking about that magnitude, we I personally have not found it necessary to then go back and run a physical analysis, you just say, look, these people are performing three times better, let's try and help more users use Zapier in a way that makes them three times more likely to stick around. Yeah,

Andrew Michael
I think that does make sense as goes back to the point that you made earlier saying that you really can't correlations here. And really, you need to test this to actually prove causation. I think that there's also another step that can be taken a little bit further when it comes to this correlation analysis. And that's potentially looking at Venn diagrams, where you not only look at the number of people that took this specific action, and then end up retaining, but then also looking at the number of people that didn't take that action and also retained. So that way, you sort of get a sense of how strong that signal is. And ideally, like what you're trying to look for is hundred percent of people took the action and retained and know people didn't do this action. And retained, just adds another level level level of granularity in terms of the signal strength and gives you a bit more confidence to run tests against,

Stephen Levin
for sure. And if you can get a Venn diagram that looks like that, you know, you really hit it out of the park with your analysis. So certainly, you want to look for that very difficult to find.

Andrew Michael
Cool. So another topic, I think that's very interesting. And I want to have a discussion and chat with you about it, we talked about it previous to starting this call is at our China, I'd imagine as awesome early at some of the companies that you work with and adapt here, we tend to have this seasonality effect, where, due to the nature of our product, people use us potentially lucky, we have a good portion of our customers that use us on a project basis. So they would have a new website upgrade, and they want to go and analyse the new site change and see other barriers. And once their projects are finished, then they use case for just stops, and they stop using the tool. And traditionally, like if you look at it, this customer could potentially be considered as a churn customer and then but what tends to happen is that we have this huge reactivation six months, 12 months, 18 months later, where customers are actually coming to the product because they have a new project again. And I'm interested to hear your thoughts on this. And what are some of the things potentially that you may be looked at? When it comes to measuring and tracking this sort of seasonal seasonal behaviour, periodic churn, as it is? I know, like a hot jar, we've had discussions but really not looked into it that much. But understanding sort of like, what is considered dormant versus what is considered churn. And I'm interested to hear your thoughts and what you've looked into around it.

Stephen Levin
Yeah, so this is a really interesting question. And it's something that even though I'm really deep in this type of analysis, and this type of data, I actually found a great write up or any sort of like specific recommendations around this. And so everyone seems to be just experimenting with it a little bit. So in that sense, I think this is somewhat of the cutting edge of SAS term calculations. And I do have a little bit of insight into what we're doing at Zapier to try and think through it. So we do have a similar case where people want to use us for their quarterly events, let's say. And so they pay for a month, they use it for even pretty heavily for that month, they just don't need it in between. And so they turn. But then we see exactly the same thing. Three months later, they come back six months later, they come back. And one of our data scientists, Chris has actually started to do some more rigorous statistical modelling to understand what is the likelihood of this user leaving after this month, or while they're dormant or turn, what is the likelihood that they're going to reactivate that actually starts to look at some of their cyclical behaviour patterns. And so of course, you can't identify this necessarily right on the first time that this user cycles. But as you start to see particular patterns and user behaviour, you can actually segment your churned user base into high likelihood of reactivation or low likelihood of reactivation using some more advanced statistical models that I am glad we have people in the team to do so that I don't have to. And you can target those people with reactivation emails or with this counter offers, or with just a reminder that, hey, you might want to give this a try. Even though it doesn't line up with your quarterly event business, you may still find benefit from using Zapier to send starred messages slack to your Gmail. So the people who are more likely to reengage deserve a bit of more of a high touch communication. And you can actually start to do some sophisticated modelling techniques to understand who those people are.

Andrew Michael
Interesting. I think this is something as well, like we just haven't had the bandwidth to get to. But I think at some point, we've also talked about looking at offering ways for customers to actually pause their accounts. To give them sort of like, I'd like to give it a three, six month pause, where we still retain data, this tool still works as is, but then they just don't access it until they need to, like more binary sort of bother coming back or not. But

Stephen Levin
yeah, and and you can see examples and in industry of the businesses who have nailed that pause behaviour, or things like Blue Apron, where you may not want your meals this week or this month, but you want to come back four weeks later. And so you just pause your subscription, or another example being Hulu, where maybe you're going on vacation. So you won't use your Hulu account for that month. But you know that you're going to want to come back later. So there are definitely examples of industry of businesses who have nailed that sort of pause and reactivate behaviour.

Andrew Michael
Yeah. And I think it's just a fascinating and I'd love to hear a little bit more from you potentially, even after the show how you went about sort of getting that predictive analysis on the different types of users that are more likely like, what are some of the variables that you're looking at to try and understand the behaviour of the likely of somebody reactivate?

Stephen Levin
Yeah, it's certainly an interesting thing to dive into later. Yeah.

Andrew Michael
Cool. So another thing I think definitely is a topic I want to bring up today being the nature of Zapier and the tool so and we probably should have done this at the beginning of the show. But for those unfamiliar with Zapier is f8 gives you sort of like an incredible power to connect different tools with one another. And it allows you to trigger different events in different tools based on behaviour. So an example might be, you could get an email with the weather, if it's hot in California. Or if I arrived home, I can send as EPA to my as up to my Google Home to switch on the air condition, or there's multiple different ways in which you can automate workflows as well within marketing and within your organisation sending specific messages or notifications to slack. And that I think the use cases are endless. Like it's incredible how much you can actually do with the tool. But I think the topic of this as well, it is an interesting one is integrations itself. So I think like Zapier has this incredible, powerful way of integrating tools. And I think, integrations I've always thought of them as a way of like embedding and building yourself into a user's workflow into companies workflow. So that that sort of gives you leverage when it comes to the cost of churning. And it's the opportunity cost of having to switch tools then becomes higher once you start having these integrations built in. And I know on a previous episode as well, we chatted to pedra from typeform. He's actually joined hot jar recently. We talked about a case study with him when it came to type four men working on integration. And we didn't dive into too much detail there. But I noticed a success. And I want to bring it up with you today. Sort of like how much of an impact do you see like integrations having on churn? And is it something that you see as well with like, the deeper people get integrated with different tools and the deepest ups, the less likely they are to turn?

Stephen Levin
Yeah, so that's exactly right. The type for Zapier turn analysis is actually published on our engineering blog. But you can see the details when they rolled out their Zapier integration. The users who use Zapier and type form have a churn rate on the order of half of the there's who use just tighten form without Zapier. So that's a major win, right? If you can embed something into your product that cuts your churn rate in half for the users to use it. That's a major major win. And so integrations do exactly what you're talking about. They make it sticky. And they make having to set all of that stuff up again, if you switch tools, just that much harder. And so while the type form one is the one that we've published publicly, we've actually done this type of study with two or three other customers that show similar results that when you can get users integrating their product with other products, the users who do that are really invested, they're really less likely your turn.

Andrew Michael
Yeah, I think it's incredibly powerful way as well of getting bested. So one thing I wanted to ask you as well is, what's one thing that you wish people would ask you that they don't when it comes to churn and retention and the metrics and tracking?

Stephen Levin
Interesting question. So something that I wish people would ask, not necessarily asked me but ask themselves is just what does churn really mean for our business? understanding where you're operating, whether that's your vertical, you're targeting users in whether that's your average contract size, whether that's your go to market and sales motion? You know, are you self service? Do you sell through a sales team? Do you sell only annual contracts, all of those things, make your churn rate feel very different. And so when you're benchmarking against other companies, one makes sure that you're talking about the same type of churn rate number to make sure that you're talking about similar types of buyer persona, or contract size, or vertical. And three, just focus on improving it like your benchmark against industry is only important, insofar as maybe it helps you raise money, or it gives you an idea of how you're doing. But realistically, it's all about improvement. And so, churn will naturally improve a little bit over time, because you have this big established base of customers that don't churn. And so as you grow, your churn rate may go down. But if you could really make a step change difference, it can have huge benefits for your company. And it doesn't matter where you started, it just matters that you're continuously improving it.

Andrew Michael
Yeah, I found it very interesting. And it's something like in terms of benchmarking you made a very good point is it's like, typically, people might hear a specific churn rates and think it's either good or bad. But it's really important to understand like your vertical, your current value that you're going after which segments of the market, because those all have big influences on what your churn rate is going to be. And like things like going off the enterprise, this SMB, you could, you will typically see drastically different churn rates, and what's good in one market may not be good in another. Exactly right. So last question I have for you today, then Stephen is sec, I want you to put you in a situation now where you've just joined a new company, you've come in and you've seen that churn is not looking good. Things are not looking rosy for the company, you've been put in charge to try to turn things around. What would be the first place you'd start trying to turn things around for the company?

Stephen Levin
Sure. So we touched on this a little bit, but it's worth doubling down on, the first thing that I would try and do is understand if turn isn't looking good. Why is it brand new users? Is it? Are you losing established users? That's much worse? Are you losing users that use a particular feature? Maybe you stop supporting it, maybe it's a brand new feature, and it doesn't work very well. So using that same chart that we were talking about the segmentation styles that we were talking about earlier, you really need to understand what's driving any sort of concerns with your turn rate. And then you can try and figure out how to fix it.

Andrew Michael
Yeah, absolutely. Just really trying to understand that have a good baseline knowledge of where you're at in order to dictate the strategies that you take to improve it. Cool. So I think it's it's been really really insightful having you on the show today, Stephen. I mean, I could keep you on for another two hours discussing the topic and going into a little bit more detail, but be respectful of your time and have the listeners. Thank you so much for joining today. It's been a pleasure having you. Is there anything you'd like to leave the listeners with? Like, how can they keep up to date with your work and

Stephen Levin
Sure, so I do want to say thank you for having me on. It's really been fun. You can find me sometimes writing on the Zapier blog, dap, I, er calm or my own blog at Stephen Levin dot CEO.

Andrew Michael
Awesome. Well, thanks very much for joining us today and have a great day and a good week.

Stephen Levin
You too. Thanks. Cheers.

‍