Driving user engagement for Vookmark through data-driven notifications
Vookmark was an application that enabled users to bookmark videos across multiple sources to watch later on a browser, AppleTV, iOS or Android device. Since it was an in-house application, it provided ample access to server data to enabling actionable insights on user behavior and user engagement
Key tasks a user could perform on the app
Bookmark an interesting video to watch later
Watch the video
A few primary metrics (besides DAU)
Exploring the data
Observing week-on-week growth rate helped explore possible drivers of user acquisition. The surge in growth rate (Shown in above graph) coincided with the product featuring on a tech magazine with a large reader base
Understanding patterns in bookmarking behavior helped obtain actionable insights to drive engagement through push notifications
Predominant bookmarking behaviour was gauged by an understanding of the type of extensions videos were bookmarked from, types of sources of videos that were bookmarked, days of the week where bookmarking was highest, hours of the day where bookmarking was highest. Further hypotheses were analysed based on preliminary findings
The ideal flow doesn't just require the user to bookmark videos. It also requires the user to view bookmarked videos later. Predominant viewing behaviour was gauged by an understanding of the type of platforms from where bookmarked videos were watched, hours of the day where users predominantly watched bookmarked videos. An additional metric was the average no. of days it took for the user to first watch a bookmarked video. Preliminary findings helped shape more hypotheses and explorations.
Insights on content of the videos being bookmarked and viewed helped understand the type of content that's more likely to be bookmarked. Possible experiment: Placing future advertisements in the midst of such content would likely attract the right user base. We built a word cloud to analyse the types of content being bookmarked (word cloud below is for illustration purpose only. actual word cloud cannot be repurposed owing to product-confidential insights)
Understanding the average duration of videos being bookmarked helped us gauge if the watching was more likely to be more experience-driven (example: Netflix) or on-the-run.
Hypothesis: Longer duration videos are more likely to benefit from 'experience-driven' viewing than shorter videos that are watched on the run.
The findings helped us determine if the application's viewing feature needed to expand to Smart-TV devices
User engagement roadmap
Analytical insights on user behaviour and extent of completion of key tasks governed the push notification roadmap. Push notifications for users were clustered by the stage the user belonged to: onboarding for early users, habit-forming for intermediate users and rating and sustained engagement for users who'd made a habit of key tasks.
Outcome: Leverage insights from the exploratory data analysis resulted in over a 150% increase in the average watch-ratio. The direct access to server data (perks of an in-house product) expanded the analytical scope, helping us drive user engagement while making strategic decisions on the product roadmap.