Startups | 6 min read

Churn, Retention, and Reengaging Customers


Since we launched automated emails in Intercom, one of the most common use cases has been re-engaging customers who have stopped using a product. Let’s look at how to do that effectively.

Churn Is Always Ambiguous

Churn is a badly defined concept at the best of times. App owners talking about churn percentages ignore too many variables.

When someone tells me they have 5% churn and asks if that’s good, I can never answer easily. It’s such a vague figure that I have to fall back to a set of clarifying questions:

  1. Does this figure include users who signed up during this month, or only lost customers from the previous month? In short, if we started with 100 customers, gained 30 sign ups and finished the month with 110 customers, what churn rate would we ascribe?
  2. What is the revenue churn compared to the user churn? 5% is fine if it’s only free users you’re losing, but what if that 5% happened to be all your premium customers, or 50% of your revenue?
  3. What is the activity churn? How many customers were previously active but have now stopped using it? This is usually an indicator of churn to come.

As you’d guess, these questions keep coming until we get some real insight. Aggregate figure, whether conversion rates or churn rates rarely offers any substantial insight. As I’ve said many times, in the absence of any meaningful data or metrics, show me your bank statement.

Activity Churn Is What Matters

Activity churn is where the rubber hits the road. Typical Churn stats use account cancellations as a measurement but cancellation is only ever a trailing indicator. It’s the last thing that happens.

Customers don’t make a snap decision to quit an app and delete all their data on the spot unless the app has really screwed up for them. Typically customers gradually stop using it, it goes from every morning to every week to once a month. Eventually their data in the app grows out of date and soon they realise they’re paying $29 per month for something they don’t need and don’t use. At this point you can’t recover them easily, but this is the first time a churn metric will recognise them.

Addressing Activity Churn

The most effective time to address churn is when you see a drop in usage across an entire team. This is where many auto-mailers get it wrong. I’ve come back from holidays to find dozens of “We Miss You, Please Come Back” mails, which are misplaced. It’s acceptable for one user from a team to disappear temporarily; it’s when the activity of an entire team is dropping that you worry you’re losing a customer.

When you identify previously active customers who are slipping, a well-timed personal email can help to re-engage them. The best case is that you recover a customer; the worst case is often that learn why you users leave. Either leaves you in a better position than doing nothing.

Guidelines for Re-Engagement Mails

This is a relatively new type of email, though as with all communications the guidelines will sound familiar.

1. Target the right customers

There is a difference between someone who didn’t convert after a 30 day trial and a year long user who hasn’t started to slip away. The difference results in two different jobs, conversion and recovery. Don’t let them both fall into a naive “we miss you” bucket. Treat them differently.

2. Be Personal

Don’t auto-mail a customer who has several open support issues to remind them to login and don’t begin your mail with “Dear Customer”. Such activities do more harm than good. Introduce yourself in the email, and make it clear that you’re a real person who really wants to hear what they have to say about your application.

3. Be Interesting

Trying to be cute and saying “We miss you, please log in” doesn’t work. You have to motivate a user to log in. There are often some features that will bring customers back, or prevent them from switching. By offering churning customers a glimpse of what’s coming down the line, you can excite them about future releases. Things that inspire people to stick around are usually features which save time (e.g. better importing), increase efficiency (e.g. integrating with 3rd parties), or offer additional value for no extra work (e.g. weekly reports).

If you have long standing problems that are well known (for example: performance, accuracy, limitations, etc.), it’s tempting to say they’ll be addressed, but the chances are you have been saying this for months if not years. In this cases if you can’t make a concrete promise then it’s better to say nothing, and focus your efforts on resolving the problem at hand.

4. Be Visual

If you have new features or even UI refreshes coming down the line, screenshots can be particularly interesting for long standing users. If the improvements are in terms of workflow, sometimes a clear diagram will be seen and understood far better than paragraphs of text. Never forget you’re just one of many emails in the inbox, and they’re all making promises. Any effort that makes your email stand out will help.

5. Be Thankful, Honest, and Respectful

Some amount of churn is natural. Businesses come and go, as does demand for your product. When a customer decides to leave, make sure they leave on good terms. If they have legitimate problems with your product, acknowledge them. Don’t delude yourself and fight back. Thank them for their custom and let them go. The easiest way to screw this up is to continue to spam them months or years into their departure. It does more harm than good.

The Impact Of Churn

A post by Chaotic Flow revealed some interesting data points on churn (though be warned: your mileage is guaranteed to vary). Companies with less than $10M in annual revenue have a median churn rate of 20%. Unsurprisingly, companies earning more revenue see this figure drop to a more respectable 8.5%. This isn’t too shocking. In early days companies are still finding their customers and will inevitably attract the wrong customers for whom their product isn’t a good fit, hence churn should decrease for start-ups as their product and brand becomes more focused and mature.

It’s difficult for start-ups to assess the real impact of churn as they lack the longitudinal data. Churn is usually seen as a factor limiting growth, but the impact on customer lifetime value (CLTV) is more significant. The entire SaaS business model assumes a low spend per month, but with many months revenue. If churn is significant, then CLTV is too low to support a SaaS model. This is common for products that immediate upfront value, but give no reason to stick around months later. In these cases, you can spam customers all day long but you’ll see a better return from rethinking your business model.

As always, if analyzing churn, cohorts, or life-time value gives you results that you didn’t expect, then you should be talking with your customers more often to work out what’s going right and wrong.

These days you can automate and visualise almost anything, except caring. That has to come from you.