Making Sense of Customers who Cancel

A behavior design approach to consumer SaaS churn

If there’s one number that indicates the health of a company with a recurring revenue model, it’s user churn. Whether in consumer or enterprise companies, the rate you lose customers tells you volumes about how important your product is to people. Churn represents not only sunk customer acquisition costs and lost revenue, but more importantly, unhappy customers.

There are several macro strategies for disrupting churn, such as preventing customer cancellations, resurrecting cancelled customers, and upselling current customers (known as negative churn). Before pursuing one or all of these strategies, though, get a firm grip on why customers cancel.

Hypothetical case study: Audible is a books-on-tape subscription service provided by Amazon. Audible customers onboard with a 30-day free trial, after which, the cost is $14.95 per month. For that price, customers can download ad-free, celebrity narrated books to listen to on or offline. Books purchased can be kept forever, even after the subscription ends. Customers can listen on any device, as long as the Audible app is downloaded. Audible has a current churn rate of 12%. Imagine you are a member of the Audible team tackling this challenge.

Set yourself up to dig into the churn

Step 1: Craft hypotheses for why customers cancel. List as many cancellation drivers as you can think of. Write from the customer’s point of view using your customer’s voice, such as:

• I never managed to finish an entire book.
• I could never find a book I liked.
• I didn’t like the sound of the narrator’s voice.
• I forgot I had it.
• I didn’t use it enough to justify the monthly fee.

Step 2: Examine analytics for behavioral patterns. Surface whatever quantitative data you have that reveals which / when / how customers cancel. You know 12% of customers cancel each month. But if you group customers based on the month they signed up you might see that customers within 90 days of onboarding are churning at a rate of 12% per month, and once they get past 90 days, they churn at a rate of 5% per month. Maybe tablet listeners churn more than smartphone listeners. Or perhaps customers who listen four times per month churn less than those who listen once a month.

Dive into other quantitative data — numbers like conversion rate, scroll depth, and heat maps — that might indicate when something is wrong. Look at funnel analytics. How might the length of a listening session within a customer’s first week correlate to retention?

Examine strategic partner data too, if applicable. Are there differences in churn between small and large company customers? If Audible has partnerships with universities, for instance, how do those customers compare to someone who is a single, lone customer? Or do subscribers from a particular vertical/lead source/etc. have a tendency to churn more?

Step 3: Identify who to learn from. Zoom in on cohort analytics. This will provide some insight into which customers to focus on. There is a difference between someone who quits within the first 30–90 days and a year-long user who has started to slip away. These customers must be treated differently. An easy way to define cohorts is according to frequency of use or date of sign up. Here is an example cohort analysis spreadsheet.

Dig in with qualitative research

Design and deploy qualitative research to document customer stories. If there is a customer support person or team in tact, be sure to start there. Are there already obvious reasons people cancel? If not, craft one or two questions customer support can regularly ask when they have time with customers, then look for patterns as those data points come in.

A quick and inexpensive way to solicit feedback is with a one-click open ended question. Send a simple email survey to ask “Why did you cancel?” Or build it into your cancellation user flow so when a customer unsubscribes, she can tell you why. Be sure the response form is an open text box.

Another fairly inexpensive way to learn is to talk with customers over the phone. Recruit recently cancelled customers via e-mail for a 20-minute interview for which you pay them (a $30 gift card will suffice). This will give you a deeper understanding of their main problems. Ask exploratory questions like:

• How did you first learn about Audible?
• What triggered you to sign up for the trial?
• What was the first book you listened to? Did you listen to the whole book? Why or why not?
• Which device(s) did you use to listen with Audible? Which do you prefer and why?
• How often did you use Audible? Always at the same time of day?
• What were you doing while you listened?
• How long did you subscribe to Audible?
• Why did you choose to listen with Audible versus other books-on-tape services?
• How might you describe Audible to another person?
• Where are you listening to books/stories now, if not with Audible?
• Are other companies offering you discounts/promotions that triggered you to switch? If so, which ones?
• Can you imagine a scenario in which you start using Audible again?

Once you have qualitative data, see how and where it matches to quantitative data. Do the stories shed light on the patterns? How much of your customer feedback matches your hypotheses? Which cancellations drivers are the most common? Which drivers recur per cohort?

Look to see what themes emerge from your data. Because you’re at a consumer SaaS company, cancellation themes probably look something like this:

• Nothing triggered the customer to use our service;
• Finding and listening to a book was too hard;
• Using our service was not enjoyable enough;
• The customer never developed a listening habit.

At this point you might think, “Wait, we totally have awesome call-to-action designs” or “We make it so easy for our customers to listen.” But your churn data is telling you otherwise. This is when behavioral science can further explain.

Better understand your data with behavioral science

One of the most useful behavioral models for analyzing consumer churn is the Hooked model by Nir Eyal. A Hook is an experience designed to connect a customer to your service with enough frequency to form a habit. Companies like Audible need their customers to develop a habit as soon as possible because frequency of use is a prerequisite for subscription service success.

Nir Eyal’s Hooked Model

Every Hook begins with a trigger. Humans need a cue to act and a trigger is a cue to action such as a growling stomach that tells you to eat. Or a red traffic light that tells you to stop the car. Whether it’s an external trigger like a light or an internal trigger like an emotion, effective triggers prompt immediate action. For an Audible customer, it might be a pop-up message that says “listen now” or a feeling of stress during the commute home.

If, however, your customer tells you “I forgot I had it” and “I’m just not used to using an app for e-books” then she was never effectively triggered to use your service.

When a customer is successfully triggered she then acts. Action is the second step of the Hook. Action is the simplest behavior done in anticipation of a reward, like scrolling or pressing play, and is best outlined by The Fogg Behavior Model. An Audible customer must be able to search for, downlaod, and play a book with as much ease as possible.

If that customer instead says “I could never figure out how to get a book I liked” and “I never got through an entire book” then the design for action is too complicated.

Once a customer acts, she then experiences the third step of the Hook, the reward. A reward is when your customer gets what she paid for. Audible customers may want to listen because it relaxes them or they enjoy the sound of the narrator’s voice.

If your customer tells you “I did not enjoy any of the stories I listened to” and “I didn’t like the sound of the narrator’s voice” then she never experienced a reward for using your service.

The final step of the Hook is the investment. An investment is about storing value. It’s when a customer does something within your service in anticipation of a future reward. For an Audible customer, an investment would be downloading a new book so it’s ready to go when she shows up to listen. And the more books that customer downloads, the better Audible can serve her preferences. Investment is important because it increases the liklihood of using the service again.

If your customer says “Every time I opened the app I never had a book to listen to” or “I wish the service would suggest books to me” then she’s failing to invest in your service.

At the end of the day, your customer cancelled because she never formed a habit with your service. This is why both qualitative and quantitative data are important for churn strategies: because the combination of the two will validate why and when you are losing customers. You can pinpoint which part of the habit formation journey she quit. And then you can decide what to do about it.

*Thank you to Max Ogles, Nir Eyal, Emily Goligoski, Dan Ledger, and David Habif for weighing in on this post.

*The author has never done any formal work with Amazon.