At Canva, our mission is to empower everyone to design anything. Over the years, the definition of “anything” has expanded: from social media mate

Understanding a Diverse User Base with Frequency Segmentation at Scale

submited by
Style Pass
2023-03-17 03:00:05

At Canva, our mission is to empower everyone to design anything. Over the years, the definition of “anything” has expanded: from social media material, resumes, and cards, to infographics, presentations, print, and video. Canva’s user base is also diverse, with widely varying needs and usage patterns. We have marketers who make multiple social media posts per week, and casual users who only return every few months when they need to design a birthday card or invitation.

Canva continues to expand, most recently into the productivity space (see our new worksuite launches from Canva Create 2022), and we hope that people can fulfill even more design needs through Canva and return to our product more frequently.

In this blog post, we’ll walk through how Data Analytics is helping to steer this company’s strategy. First, we’ll cover how we segment users on a spectrum of less to more frequent, and how we account for inactivity. We’ll then explain how we operationalized these segments, and some exciting applications we see across the company.

To quantify the opportunity for nudging users towards a more frequent need, and to track whether we were successfully doing this over time, we needed to assign frequency buckets to each user. That is, which users typically use Canva weekly, monthly, or longer? Looking at existing research in this space, we came across Buy-Till-You-Die (BTYD) models. BTYD models are used in non-subscription settings (for example, someone purchasing from an e-commerce shop) to predict the number of purchases a customer will make in the future given their past transactions. For us, each day a user is active can be thought of as a purchase, and we’re interested in estimating how often the user will be active (purchase) in the future.

Leave a Comment