The counter-intuitive guide to retention with Andrew Michael from Churn FM

I think probably one of the most impactful things you can do is identifying your ideal customer profile (ICP) and then just focusing on acquiring them. It starts from there because everything downstream becomes a problem if you don’t. Your product decisions, your activation and onboarding, and if you don’t have a really good understanding of who your ideal customer profile is and what their jobs to be done are, you can’t do anything else right. You’re just winging it, guessing what you’re building, and throwing things at the wall.

Most early-stage founders don’t want to do this; they say, “who’s your customer? Everyone!” They don’t want to negate a large percentage of their total addressable market. And in the early days, you do need that kindling to get the fire going. But when you start to get to a decent customer base, you really need to double down and focus on who is using your product the best. Are they using it in the way you believe it should be used? Is this the direction you want to take? There needs to be a balance. This also extends throughout the business, and you need to incentivise your sales team not just for closing deals, but for closing the right deals, perhaps with commissions based on retention, not just the initial sale.

Almost 99% of people, when they see a churn problem, their natural and first instinct is to put in a churn exit survey. I think that’s a completely wrong place to start. The place you should always start is understanding who is successful with your product. What did they do in their onboarding experience that led them to become successful? Who are they, and what do they do that makes them successful? Then, focus on finding more of those people and helping more new users get to those same actions.

Activation is always the way to go because it has this compounding impact over time, rather than focusing on the reason for churn at the end of the journey. It’s prevention versus cure. When you focus on the churning users, you’re often getting feedback from people who were probably never going to be good customers anyway. You’re better off asking your successful customers, “what nearly stopped you from renewing today?” and gathering feedback from those who were maybe on the cusp of churning but didn’t. Then you can go out and focus on fixing those things.

The “customer champion” is a big one that comes up frequently. You need to know when a customer champion leaves their company because the risk of churn becomes very high for that account. This is the person who was internally fighting for your tool, sharing updates, and showing its value. When they leave, it creates a void where people start asking: “Why do we use this tool? What’s it for?” If you don’t have a good network within that account as a backup, there’s a big risk of churn. I remember an interesting interview with Julian Shapiro, who almost built a product off the back of this idea because it’s such a critical signal.

There’s a very basic formula to calculate your “growth ceiling,” which is based on your current growth rate, customer number, and churn rate. You can literally calculate in which month you’re going to stop growing unless you move one of those levers. I think this is a very good exercise for everybody to do. We did this at one company and realized, “Oh shoot, we have 12 or 18 months until we stop growing; we better fix this.” That forces you to either expand growth channels or, more importantly, reduce churn. When you fix both, you can hit an exponential curve.

This came up in an interview and challenges the idea that you need to get people to value as fast as possible. Sometimes it’s better to delay that gratification and focus on getting things set up correctly. This actually came from an interview with Eleanor Dorfman from Segment. Segment’s value proposition was that you could get clean tracking data and send it to all your downstream tools. In the beginning, they just let people get started on their own right away. What they found was that people were very quickly making the same mistakes they did in the past, just having junk data in all their tools instead of just one.

So, they implemented a step where they forced customers to create a tracking plan before they could use the tool. This is very counter-intuitive; it’s a big blocker in the journey to getting started. But what they found was that it increased retention tenfold and engagement in the product because people weren’t getting to those pain points again. Superhuman did the same thing with their onboarding, forcing users to book calls. This idea of getting to the “aha moment” is important, but you need to make sure that moment is tied to long-term retention, not just a short-term win that’s going to hurt people later.

There were a number of things Hotjar really excelled at. The first was an insanely high talent bar in the early days. In my earlier companies, while we hired smart people who went on to do great things at Google and other startups, when we hired them, they were a lot more junior and needed time to learn. Hotjar didn’t hire juniors; they only hired senior people in their roles in the early days. This makes a big difference in setting the bar where it needs to be. It was also important because we were a remote business, and it’s harder for junior people to learn in that environment.

I’d also previously raised venture capital twice, so there’s a different level of focus on metrics. In VC-backed startups, at least at that point, the focus was a lot more on driving growth and user numbers. But at Hotjar, being a bootstrapped business, we had a maniacal focus on the details and on growth through sustainability. We were focused on what the revenue per head was, the LTV to CAC ratio, metrics that have become important now but weren’t talked about as much back then. We were ahead of the curve on how to build a sustainable business.

Ironically, when I joined Hotjar, we were barely using data, almost zero. We had a Mixpanel setup, but it was so bad that nobody trusted it. I was shocked. I came from a VC-backed world where investors wanted every metric under the sun, so I was focused on having the most robust analytics. In retrospect, I think Hotjar’s approach was the right way to go at the start. In an early-stage startup, your data can be skewed very easily; you don’t have enough volume for statistical significance and you get a lot of false signals.

We only really had a robust analytics setup in year four or five. In the beginning, growth was driven by user feedback and founder intuition, which went against the “test and iterate” startup mantra at the time. You need some data for signals, but most of the time, those signals aren’t going to be as strong as actually speaking to ten users and gathering their feedback.

It was the first time I had worked in an environment where people were not just further along than me in their journey, but were also just so much more talented than me. It was a good learning space where I felt, “Okay, there are people here that I have a lot to learn from.” I wouldn’t point to one massive thing but rather a collection of small little details that together are incredibly powerful.

For example, a learning that came from our content team completely changed my approach to landing pages. My previous approach was to start with a layout and design and then try to fit copy into it. They were adamant: “No, this is the absolute worst place to start. Focus on what you want to say first, and then we’ll figure out the design.” It’s “form follows content.” I used to obsess over making everything look great, but I learned it’s way more important to have a really strong user experience than a fancy, shiny UI. I remember joining Hotjar and we had one of the ugliest products, but it was one of the most effective.

Another learning came from our CEO, David Darmanin, who had a strong background in conversion rate optimisation. He taught us the importance of when to collect feedback, using a Star Wars analogy. He said you have three types of visitors: Stormtroopers, who will always be customers no matter what; wondering Chewbaccas, who are 50/50 on converting; and Jedis, who were never meant to be customers at all. The problem with asking for feedback on your landing page is that you get it from everybody, including the Jedis, which gives you mixed signals. You’re far better off asking immediately after purchase, “What nearly stopped you from converting today?” because you’re getting feedback from your ideal customers, the Stormtroopers and the converting Chewbaccas, and you can focus on fixing things that will raise conversions for them.

I was at Hotjar at the time, maybe a year or two in, and I wanted to start something new but wasn’t sure what. I was talking to Louis Grenier, host of “Everyone Hates Marketers,” and he said, “If you want to get started, build an audience first, and then when you’re ready, you can sell to that audience.” That’s where the idea came from.

I wanted to start a podcast for two reasons: one, I could learn, and two, I could start building an audience for whatever came next. I chose churn and retention specifically because it had been a bother for me in the past with the businesses I’d built, and it was also something I was responsible for at Hotjar. I was tired of reading blog posts like, “all you need is five friends in seven days,” where everyone thought they were Facebook and could find one magic number to fix their retention problems. I knew there was no one-size-fits-all solution; it depends on your business, stage of growth, and customer segments.

I also knew that if you asked any CEO what they would be willing to pay to solve churn, they’d probably say, “whatever’s in the bank,” because they know it pays back in compounding interest. If you’re running a SaaS business and people are canceling, you don’t really have a subscription business. Also, at the time, it wasn’t talked about that much. The focus was all on VC-fueled growth. Then, just after COVID, the crash came, companies hit their growth ceilings, and suddenly churn became the problem everyone was talking about.

The industry has changed entirely, even in just the last year. I don’t think software itself, the product you build, matters as much anymore. What’s more important is distribution, brand, and user experience. I would only build something now where I felt I had an unfair advantage in distribution. Even if you come up with the most novel software idea, it’s just a few prompts away for somebody else to copy it. You see it in most industries now, this convergence where everybody is building everybody else’s features. So if the features aren’t the differentiator anymore, you need to find your key differentiator somewhere else.

Hotjar is a good example. They did an incredible job on brand, building a narrative in the startup community and democratizing access by being so much cheaper than alternatives. They became the de facto brand people trusted because of the messaging and culture. If you look at features, competitors offered the same thing, but the warm, friendly brand was the differentiator.

A good example on the flip side is PostHog. When they entered the market, they realized software was becoming a commodity, so they decided to treat it that way. They went with a cost-plus-margin pricing model to be the absolute cheapest in the market. Everyone else was still trying to do value-based pricing, but PostHog saw the future and decided to compete differently.

There are obvious use cases where you can increase retention with AI. The first is personalization. In the past, it was almost impossible to personalize on a one-to-one basis. Now, you can use AI to create custom summaries of a user’s activity and give them unique insights. Instead of an email saying, “you had 5,000 visitors this week,” you can now say, “you had 5,000 visitors, and interestingly, I noticed 50% came from Brazil…” You can go so much further. This is also hugely impactful for high-touch customer success; a CSM can have an LLM do deep research before a call to build instant rapport.

Products are also getting better, faster. Think about bug reports. In the past, a bug had to be reported, put in a backlog, prioritized, assigned, and then worked on. We’re almost at a stage now where somebody can report a bug, it can go to an LLM, the LLM can search the code base, find the bug, implement a fix, and push a pull request for an engineer to simply review and approve. I was talking with a friend who is building a feedback tool that connects to GitHub to do just this. The speed of product improvements and bug fixes is going to have an exponential impact on retention. If you can solve people’s problems instantly, there’s almost no reason for them to churn. You could end up with the first truly self-improving product.

To be honest, there aren’t many great books written specifically on churn and retention; pretty much all of them have disappointed me. But if I’m forced to choose one, it’s probably Delivering Happiness by Tony Hsieh. The focus on customer experience is key; if you can really nail that, you can nail retention as well.

My favorite app on my phone is probably a tie between Strava and Twitter/X.

I’m an extreme person; I live my life to the extremes on every end, from work to life to family. I’m trying to maximize every hour of the day. For example, I’m currently doing my own version of the “75 Hard” challenge. For 75 days straight, you have to follow strict rules: no junk food, drink four liters of water, exercise twice a day, read for 15 minutes, and if you miss a day, you start over from day one.

I’m doing that, plus I’m learning to trade, so I have about another 20 rules I need to follow every day for my trading. My goal is to be extremely disciplined, follow a process, and make 75,000 in 75 days. I’m even recording a YouTube series about it. It’s about building incredible discipline. Once I get locked into something, it’s done. I’m not going to miss a day.