The DataHero Blog

Estimating Customer Churn in DataHero

March 18th, 2014

DataHero_EstimatingCustomerChurn

Customer churn is a metric that essentially tells you how many of your customers are satisfied enough with your product to continue paying for it. It’s an important metric to monitor the health of your overall business model and aid in predictions about future revenue. Unfortunately, this number can be measured in a number of different ways and many online payment platforms don’t make it very easy to calculate.

DataHero offers a few different ways to get a pulse on your customer churn. In this post I’ll outline three charts that will allow you to estimate churn in DataHero using Stripe, the online payment platform. These are some “hacks” you can perform that will add some information to the data we already pull in from Stripe and will provide a churn percentage.

Get Started

Connect your Stripe account and import the Customers and Invoices report.

Cancelled Customers

Step 1: Create a New Chart

Step 2: Drag on the following attributes:

  • Cancel Subscription at Period End
  • Subscription Start

Step 3: Click the Percentage button in the top left corner of the chart.

Because your data contains a true/false attribute, DataHero automatically creates a stacked chart that allows you to visualize how many of your new registrations cancel by the period end date. By clicking the percentage button, you can then see the percent of users who have cancelled during a given time.

Adjust the time grouping on the left-hand bar to view subscriptions and cancellations by day, week or even day of week.

It’s important to note that you can set your period end date in Stripe. For our model, we bill monthly so the period end date is one month from when a user registered. Match your Stripe billing period to the time grouping period in the DataHero dropdown menu for an accurate count of customers.

Drag a specific attribute onto the bottom portion of the chart where it says “Create a Filter” to visualize only cancellations, for example.

Ensure that only “True” is selected from the menu – meaning that the subscription was indeed cancelled at the end of the period.

The resulting chart displays your cancellations by day of the week.

Cohort Analysis of Transactions by Customer Creation Date

Cohort analysis allows you visualize how groups of users behave based on when they created their accounts. As mentioned before, churn can be calculated in a number of different ways and has many variations from a statistical perspective. Cohort analysis, on the other hand, allows you to get an idea of how many customers are dropping off their plans after they’ve created their accounts.

Step 1: Create a New Chart

Step 2: Drag on the following attributes:

  • Invoice Date
  • Customer Creation Date

Step 3: Click on the “Start Editing Chart” button and ensure that above “Customer Created” you see “Number of Records”

You’ll see the resulting chart of customer transactions based on when the customer was created. If a user created an account in January, he/she is a part of the January cohort and you can see how many subsequent transactions occur from this cohort each month after January. I’ve included three examples below that will help you understand three different churn scenarios. In these scenarios we will acquire around 100 new customers every month. This allows us to focus on what the cohort drop off (churn) shapes may look like. Your monthly cohort sizes and behaviors may vary by season but ideally will look more like a hockey stick with a month-over-month increase.

The first churn scenario would occur if you have a high churn rate.

In the scenario above, almost as soon as users register, they cancel their accounts and have no recurring purchases.

In this second scenario, we see a more gradual churn rate but we do see that eventually all users drop off. The long-term benefit of the service may not be clearly communicated to these users.

In this final churn scenario, we see that there is still some churn month to month, but it eventually stabilizes and retention does not drop to 0.

If we visualize this data in a different way (change the chart type to a Stacked Area chart on each cohort scenario) we can visualize the long term effects of churn rate on our customer base. The following dashboard reflects the three different scenarios in terms of overall transactions by customer creation date cohort.

The Stable Churn scenario is obviously much more beneficial for our overall business in customer base and in number of transactions.

Change the time groupings on the Customer Creation date or the Invoice Date to month, quarter, or week, by clicking the dropdown menu below each attribute.

Using these connected payment services, you should be able to estimate your customer churn in a few clicks. If you use a payment service like PayPal, Zuora, etc. outside Stripe, simply export the data from your online payment service to a CSV and upload that into DataHero. Let us know what other churn questions you have via email or in the comments below.


DataHero helps you unmask the answers in your data. There’s nothing to download or install. Simply create an account and connect to the data services you use everyday (like Salesforce, Stripe, MailChimp, Dropbox and Box). DataHero automatically decodes your data and shows you the answers you need through dynamic visualizations.

Visualize My Company’s Churn

By Kelli Simpson

Create my Free DataHero Account

Get the fastest, easiest way to understand your data today.