Hello Sunday Morning Case Study - Reducing Alcohol Use
Alcohol and other Drugs, Digital Health, Data Analysis, Machine Learning
Daybreak: Peer support for reducing or quitting alcohol in a digital app
The Daybreak App
Sample of clustering analysis of App usage and engagement for 30,000 clients
THE SOCIAL ISSUE
Across Australia, hundreds of thousands of people struggle to get their drinking under control, impacting their health, employment opportunities, financial wellbeing and personal relationships. Launched in 2016, the Daybreak app was developed by Hello Sunday Morning to facilitate peer-to-peer support to help participants reduce or quit drinking.
THE CLIENT’S CHALLENGE
Hello Sunday Morning (HSM), a digital health organisation, was founded in 2009 to provide online resources and opportunities for people to record and track their health improvement. The Daybreak app helps Australians to change their relationship with alcohol with access to an online peer community, habit changing activities and supports, and one-on-one chats with 'care navigators'.
Over the past two years, more than 47,000 people have downloaded the app and while data suggests that some drop off early, many users see significant benefits. After years of app usage, HSM had millions of data points but were yet to fully make sense of the data and understand the stories it was telling them. They wanted to know how they could continue to improve their impact, how to generate high levels of engagement via the app, and how to drive even higher improvements in alcohol usage and harm reduction.
THE ROLE LATITUDE NETWORK PLAYED
Latitude Network collated all the de-identified usage data so that no individual could be identified (no names, emails etc.), but also we had no access to any content the user posted to ensure full privacy. We were interested in the overall patterns of usage, not any individual’s actions. We developed new metrics that gave HSM staff a better insight into engagement and actions on the app (e.g. how many times someone might pick up the app over a time period and many other metrics). We then used Machine Learning tools to identify natural clusters of usage and engagement, as well as clusters of behaviours such as type of posting, frequency and intensity of posting and numbers of actions taken. The aim was to identify how participants were using the app and the opportunities for improvement and change.
<-- The graph to the left illustrates just one of the outcomes of this deep-dive into HSM’s data, identifying five different types of activity behaviour (the data is disguised). Being able to segment app users into separate categories and behaviour types helps identify different needs and behaviours as well as those who could benefit from additional or different services.
THE TRANSFORMATION AND KEY OUTCOMES
Thanks to the in-depth data analysis we carried out, Hello Sunday Morning now has a range of rich, detailed information about their users and their usage patterns. One group is those people who are active on the app but could benefit from a mentor to support them to take on a greater role in the online community. Another segment identified is people who engage over long periods of time but at low levels. Insights such as these have helped HSM consider ways they could prompt various cohorts of users to become more engaged, reduce drop-off rates or increase uptake.
Overall, the data analysis is helping HSM to identify strategies to increase engagement and better measure outcomes, while also providing them with tools and processes for collecting new data that will demonstrate the app’s effectiveness across different segments.
This ability to segment app users, better understand what the data is telling them, and further analyse and collect data is providing Hello Sunday Morning with richer data that tells a story, a story they can use to further reduce alcohol harm across Australian communities and target increased funding.