How To Systematically Fight SaaS Churn and Win – Robert Graham – MicroConf 2015

0:16So my name is Robert Graham. We're going to talk about fighting churn and winning. Uh one thing I'd like to do first though cuz the the timer doesn't start till I advance the slide is uh is Rob and really everybody here, uh I wouldn't be up on this stage and talking about business unless I had come to MicroConf. And I don't know if I have time for applause or anything, but I just like to tell you all everybody in the community, especially Rob and Mike,

0:38thanks.

0:44So we're going to jump in with $66,000. Uh so I did a best effort estimate. I have some experience with businesses at MicroConf. This is what I think you're losing this year if your churn is worse than average instead of average or better. This is an annualized number uh and it's a lot of money. So we're going to talk about how you can use decades of research cuz there are other businesses that are subscription-based and they've already done this research and how you can use that to look at churn in a new

1:11way. Uh something that I came up with, called an activation map, can help you identify where your churn problems are. But you can call it a lot of things. Uh and then one really tricky way that the worst solution you find to a problem can appear to be the best solution. We're going to talk about how to fix that. This is my favorite MicroConf tradition

1:31by far. Uh these are my children. Uh again, my name is Robert Graham. Uh I did two products in wildlife management. Some of you may know me as the white-tailed deer guy. I wrote a book called Calling Early Customers, helps engineers uh make cold calls. And now we're working on Keepify, which helps businesses reduce their churn. Um and first we're going to talk about what churn is just to get started. So this could be a talk by itself, but we're going to go with simple churn. So, simple churn is just if you have 100 customers in April and you lose 10 of

2:02them, you have 10% churn. Let's do an example to get a better sense of what this costs. So, if you had 500 users paying you $49 a month and you were growing at 12% month over month, what does it look like if you have average churn or worse than average churn? So, there are no numbers on the graph except for 5 and 10%. So, the blue is 5% churn and uh sorry, the blue is 10% churn and what your revenue looks like and the orange is if you were down at 5% and in the grand tradition of leading questions and slides, how many of you think this is

2:35more than say $100,000 that it costs you? Just like one guy? It was a leading question. Oh, man. Well, you're all right. It's $88,000. Uh so, but one thing that this number, this $88,000 number ignores is is LTV and how much it costs you to acquire those customers. So, this is also something that can change your business to sign a direct loss. If you have 5% churn, then you're almost $1,000 in LTV per user, whereas if you're at 10% churn, you're down at about 490. So, how much you can spend to profitably acquire someone is vastly different if you bring

3:17your churn down. Um and so, what we're going to focus on today to fix our churn problems is customer engagement. This is both because I think it's underserved as a thing that you can do to reduce churn uh and because there's some excellent talks on some of these other subjects. So, the first thing to do with uh

3:36customer engagement is mindset. Uh a lot of people can come to different parts of online businesses just with a kind of a funnel mindset. And there's one problem with that in churn and that's a lot of times churn represents someone's loss of faith in your business. it's that they don't believe in you anymore, and coming to them at that

3:56point isn't very productive. So, what we want to do is make those people successful when they get started, and make sure that they stay successful when they're in your business. Um this is me with my hands up. This is my my picture. Uh this is the top of the Harding Icefield. It's a trail near Seward, Alaska. It's a really cool place. You should go. And what I noticed when we got up to the top there is that a lot of people were turning around within like 400 yd of seeing the icefield. And it's a tough hike, but if you get within 400 yd, you might as well

4:28go to the top of that hill and see the icefield. And I think if people had been there and known that, "Hey, this is how you you know you need to walk in the snow, or these are the shoes you should wear. This is how cold it is up there. You should bring some water." that they would have been able to be successful in their journey. So, you just need a

4:44little help along the way. And this is where they would get to. So, this is customer success. This is helping people get where they want to go, see what they want to see. A really bad way to measure this is if they log into your application because logins don't represent any achievement. Logins, in fact, Buffer found that 33% of their churn was actually active the

5:10day that they churned. So, it can be just a totally backward metric. And this is where we get into research. So, all those companies, your telco, your cable company, magazines, periodicals, they've done decades of research about how to analyze subscription revenue and how to keep people around. And some of their churn problems are a

5:30lot bigger than ours. And this is how they do it. Recency of purchase, frequency of purchase and activity. And this frequency thing ties in with what Jesse said about about establishing habits. You kind of want to establish a habit of a commercial relationship with someone. And then the monetary value, so what they're spending. And the great thing about the RFM model for SaaS businesses is that our data is much more rich than any of those other

5:57businesses. So, this is where we're going to start measuring. And this is what I call the activation map. Groove uh a lot of you may be familiar with Groove. They have a great blog. And they reduced their churn with something really similar to this that they called red flag metrics by 71%. So, my example here is Drip. And I think

6:16a lot of you are familiar with Drip. It's uh it's lightweight marketing automation that doesn't suck, I believe. And uh so, on on the right-hand side is is basically what the Drip onboarding process looks like. So, you want to track that. You want to have events that makes sure people are getting onboarded. And you want to use the same as ideas to help them get onboarded. And then you also want to track deeper things. So, things that not everyone is going to do, but that you want people to do because it makes them more engaged with your application. You want people to create split tests. You want people to

6:46integrate with other services because it gives them more value and it gets them deeper in. So, we're going to measure this. We're also going to do another measurement. And it's something like happiness. But we're going to use the recency and frequency of our events to do it. There's one caveat here. I left off the onboarding events. There's a reason I did that. So, later on in the customer life, those go to zero because you don't get onboarded again. You're already a

7:12customer. But early on, if you're looking at early churn, and we'll talk about this, you may want to address that early on and measure it a little differently. The problem with these measurements comes in and it's something called Simpson's paradox. Some of you are probably already familiar with this. It's something a lot of statisticians

7:30statisticians know very well. But it's super counterintuitive. So, so bear with me. Uh you see the blue line and the red line. And they both have a pro trend, right? But these positive trend groups of data, if you combine them and make one line to represent them, it's the black dotted line, and it's going down.

7:52It's a negative trend line. This is Simpson's paradox. I'll give you another example. This is a kidney stone treatment. This is on Wikipedia. It's really famous example. Um so in in treatment group A, when you split it out into groups that had small kidney stones and large kidney stones, treatment A is better for everybody, right? But when you combine them, treatment B looks like it's the best

8:16solution. So if you didn't look at this broken out into groups, then you would be stuck with a lesser solution. It would be worse for everyone. So this is really important in medicine, and it's also really important in your business. And the thing that can save you from Simpson's paradox is really simple. Small stones and large stones, this is called a confounding variable, and you

8:35can solve it by having cohorts. Smart furniture actually increased their open rate on their emails by 311% by breaking people out from six cohorts to 30. So what types of cohorts do you want to use? The first one is calendar. So Rob spoke about early churn, and that's really important. Your 60-to-90-day churn is going to be vastly different from everything else. Depends a little bit on

8:58your business, but that's where you are. The other thing that's going to be different, because we're all startups, because we change really fast, you want to track people from month to month. How are they doing? What changed? And so people in January have a different experience than people in February. The next one, and this is this is from

9:15BidSketch. This is Ribbon's plans. Um you want to look at value. So what plan are they on or that they had? How much are they spending? And are they monthly or are they annual? Because that'll make a big difference in how they use the application, actually. The next cohort I like to do is an action-based cohort. Telefónica Europe combined this with a little bit more fancy business intelligence, but it's pretty much just action cohorts, and they were able to reduce their churn by

9:39some like wild amount. And the idea is just to know, you know, if people are collecting addresses, if people are creating split tests, what is that what are they otherwise likely to do? What are they missing? What can you encourage them to do? Where can you take them? So, how can we use all these

9:54measurements and all these cohorts? We'll talk about a couple things we can do to actually get churn down. So, Mention reduced their churn 22% overall by focusing on high-value people and using a lot of different communication channels. These are a lot of options. It actually doesn't matter to me if you want to use email, IM, or in-app stuff, text messages. There are a lot of

10:14options. You can automate all of them. And you might want to use more than one. So, let's do a couple of quick examples though. I'm going to stick with Drip, uh mostly because I just need something everyone's familiar with. So, if we have some information that tells us that people in Drip using split tests and automation tasks are you getting 20% more emails, so they're being more successful, then we want to make sure that people are doing those

10:38things. So, we could create Drip campaigns for new people that just tell them about, "Hey, these are the keys to getting started with split testing campaigns. This is how you can collect more emails with web automation." So, this is a really cool thing to do, and it'll take you a little further down the path. And another thing you can do, behavioral emails. So, triggering things. And it doesn't necessarily have

10:58to be an email, but it could be. Um so, if a user using split tests hasn't created a test in a while, send them some split test ideas. If someone creates some automation tasks, but then has just like fallen off or, you know, doesn't get started right, or maybe just to help them get started. Give them a few pitfalls. What should they avoid? How do they get going? Make sure that they get to that success they were looking for. Experian, everyone's favorite company, uh they saw a 54% improvement in uh their churn by using triggered behavioral emails. 54% is a lot for one

11:32technique. So, another idea is extended onboarding and uh this is where you steal all of Samuel's good ideas and you combine them with some of these later activities. So, you know that people say that create automation tasks do better than people that don't. So, you create some onboarding flow that's later on in the process that helps them get to that point. Now, maybe not everybody does and maybe it's an opt-in

11:57thing. Maybe it's an optional thing. But, that sense of like progress and accomplishment and bringing people along is super helpful cuz these are the people that are getting to that top of that mountain, but not getting to see the ice field. Services. This is another great place. Um I think so Zendesk did personal outreach with email and phone calls and they reduced their churn a lot that way. It's a good thing to do. Another great

12:21example is uh Brennan Dunn. Um so, I think his products and training feedback into his his products great. He does freelance software for project management, Plan scope. And then he has books and he has a newsletter that gives a lot of value to help people raise their rates and help people get more customers. People with higher rates and more customers need more projects, more project management. So, you want to have

12:42some of that synergy. The last one I'll talk about is negative churn very quickly. Uh so, this is when your upgrade revenue when you upsell people new services, new products, or just upgrading people to higher plan service. When that's greater than your downgrades and your revenue churn, you have actually got negative churn and this is this is kind of the the graph that you're looking for, the

13:03green one. Um and that's something for when you stabilize your churn and you get mature. So, when I went up there to the Harding Ice field, I didn't think that you could actually go on the ice, but I saw people out on the ice. I thought that was like environmentally, you know, not acceptable or just maybe crazy. But, I saw people on the ice and when I saw that that was possible, I wanted to be out there. I wanted to do it. I wanted

13:26go to the next level. And to me, that's customer engagement and and making customers successful. You want to take them where they can get to and you want to make them successful. So, I did uh you know, I installed a a Java application. But, after that I went uh to the Matanuska Glacier and I did some ice climbing. And I think that's what it what you want to do for your customers. You want to get them to be successful with what they need right now and you want to show them what's possible so you can make them

13:53successful. That's it. Thanks, Robert.