How to use email marketing to improve product adoption
At Mixpanel, we use our product internally to track critical business metrics. We “dogfood” it as much as we can: to measure product use at the account level with Group Analytics, onboard new customers using Mixpanel Messages, and test subject lines in our blog newsletter. But until recently, we had to rely on Marketo for some email campaigns where “dogfooding” Mixpanel wasn’t the best option.
In March of this year, we extended the use of Mixpanel to our product update emails. We wanted to experiment more deeply to drive up not just readership, but feature adoption – the ultimate goal. While opens and clicks were easy to track in Marketo, they didn’t tell the full story. With Mixpanel, we wanted to see where people go in our product after clicking the link, and what they do after. As you think through your user segmentation and email tracking, we hope our approach will inspire new insights.
Sending emails with Mixpanel
Over the past year, Mixpanel’s Messages and Experiments suite has evolved with new features including a drag-and-drop interface and advanced personalization capabilities. As we moved away from Marketo, we used these new features to create a template for our product update emails, segment our subscriber list, set up email cadence, and run a/b tests to optimize email performance – all within Mixpanel.
Targeting the right users
First and foremost, we wanted to reach the right people, not just everyone who had ever signed up. To ensure we were emailing folks that would actually care about this email, we added a filter of users who had done something in our product in the last 12 months.
The recently launched nested cohorts in Mixpanel made the setup very easy. We simply used existing cohorts to define new ones. For example, to create our “free active users” cohort, we selected users who had been active in the product, and excluded the ones in the “paid users” cohort.
Optimizing email cadence
To adjust email cadence, we added a filter that excluded users who had already received an email from us in the last 7 days.
Analyzing email performance
To measure the impact of our product update emails, we looked at how the campaign performed in terms of nudging people to go into Mixpanel to view reports. We built a funnel to see how many people opened the email and then viewed a report in Mixpanel within 5 days of first getting the message.
The analysis showed that most people open the message and come back into Mixpanel to view their reports within a day.
We also looked at our Flows report to see what people were doing after they got the message. About 17% of people went straight from getting the message to viewing reports. About 2% of users logged out of Mixpanel, and logged back in to view more reports after reading the email. A small percentage of users visited our marketing pages and read about new launches.
After running several types of analysis, we looked at the key findings:
- Users who interact with the email are more retained than those who don’t.
- Users who open the email tend to be about 5% more retained than those who don’t open the email.
- Even better, those who click on any part of the email tend to be about 20% more retained than those who opened the email.
- Email opens lead to a 20% increase in viewing reports.
While we were obviously thrilled about these results (and grateful to our loyal product update email readers), it’s important to offer the standard statistical disclaimer: Correlation doesn’t imply causation. At the same time, with Marketo, these insights simply weren’t available, and we couldn’t measure email performance at such a deep level.
Up next: dynamic personalization
Inspired by these findings, we are now implementing new changes and experiments. Our plan is to expand email targeting to reach more users, and personalize email content depending on the type of user, and which features they use.
In our next product update email, we plan to use Liquid, an open-source template language created by Shopify, to dynamically insert content into messages. For example, if we predict that you are not likely to use a certain feature, then we will not show that section to you.
We hope that these updates will help us make our product emails more engaging for you. Not on the subscriber list yet? You made it this far. You should probably sign up.