When I was at Codecademy, I would've killed to know other people's churn numbers.

We had a really good understanding of our metrics. We'd monitor them every day.

The business was growing well, but could we do better? Where should we focus? When is it time to stop working on a metric?

As the company grows, you have more and more options. More features you could build, more flows you could optimize, more experiments you could run.

The backlog gets longer. The debates get louder.

Bach then, there were no good benchmarking reports in our industry.

Thankfully that has changed.

Churnkey published two great benchmark reports in partnership with Stripe and their data.

This is real data broken down by industry and price point.

Benchmarking data is extremely useful if you are aware of the inherent tension within it:

Benchmarks tell you what's normal, not what's good.

If you're dramatically below benchmark (on the bad side), that's a clear signal.

You're probably missing best practices. Basic stuff that everyone else has figured out. Start there.

If you're at benchmark, that's not success. That's table stakes. You're average. You want to be best-in-class, not average.

Use benchmarks as an input to your planning not as the goal.

For those unfamiliar with the term, voluntary churn means people actively cancel via your cancellation flow.

This is what jumps out to me:

Voluntary Churn Across Industries

The main driver of your retention numbers is going to be how long the user's have the problem that you solve.

Travel is inherently a short-term use case, while things like leisure and insurance are inherently long-term use cases.

Companies in these industries can impact these numbers but not fundamentally change them.

Most ed tech companies would be extremely happy with 3-4% monthly churn rate, but that means you’re still losing 1 in 3 users each year.

You need a massive marketing machine to keep the growth going.

This is one of the reasons I push all startups I work with and talk to to work on their churn fundamentals early.

Plug your numbers into their calculator and do the math for yourself.

Any time I've ever seen cancellation flow data that it shows the same thing:

To me these are both the same objection, users don’t see value.

They are not willing to pay with either time or money.

These objections never go away, they are almost always #1 and #2.

I think the actual way of solving them is improving your core product and not lowering your prices.

That said, cancelation flow tactics (listed here) are very effective and Churnkey data shows this.

The more you offer a mitigating step that matches the reason the user leaves, the more effective it's going to be

These are the users that churn without actively canceling. Typically these are payment-based reasons.

Here’s what jumps out to me.

This confirmed something that I suspected for a long time

The cheaper your product is the more likely you deal with bad payment methods and accounts with low balances.

This is a hidden impact of raising your price 😀

You actually raise your ARPU/LTV in two ways

Products that inherently attract affluent face dramatically lower payment processing problems. I've seen this multiple times.

if your product caters to wealthy executives, you probably don't even know what payment churn is.

Overwhelmingly you have payment problems because customers don't have money in their account at that time.

The three big buckets of effective tactics here are:

Churnkey expands on these tactics here and I also have this post on the underlying logic here

Reason That Payment Methods Fail

There is no reason to not have these best practices set up for all web based subscription products. It makes a huge difference

If you need to convince someone, convert your monthly churn numbers to annual churn numbers.

At 5% monthly: ~46% annual loss. At 10% monthly: ~72% annual loss. At 15% monthly: ~886% annual loss.

This compounding can also work in your favor, but only if you start now.

Every month you delay is a month of users gone forever.

A final tactical point. Not nearly enough companies run cancellation flow-based A/B tests.

There are three things that impact 100% of your paying users:

Most companies run A/B tests on #1. #2 is very hard to do. Very few run them on #3

We’ll cover more on minimum detectable effect next week, but if you have 500 visitors your cancelation flow per week and you see about 300 of those people cancel, you will be able to detect a ~10% drop in 4 weeks.

This is enough firepower to run real experiments. Test different offers, different copy, different pause lengths.

This is how you build your own benchmarks and move from average to best-in-class.

Churnkey makes this easy and I would highly recommend using it.

Good luck out there.

Dan