A transparent and editable recommendation profile

Buenos Aires
4th Dec 2018
How can we give people greater control and transparency into the profiles that apps and services generate off their viewing and browsing history.
Product Context

Bizfeed is a digital media service with an emphasis on business and economic news coverage including articles and video.

In order to provide this service, Bizfeed is powered by some of the following data:

  • Bizfeed tracks which articles or videos you view to enable them to recommend things you might like and understand what you're interested in.
  • Bizfeed serves targeted advertising and pushes news from third parties based on your preferences. Alternatively, Bizfeed offers a paid add-free version.
  • Bizfeed allows visitors to comment on articles using a third party comments integration tool which allows visitors to comment by signing in with their preferred social network.

Problem & Opportunity

People want accurate recommendations. However they sometimes fall short, displaying un-relevant posts to people or misrepresenting them. How can we then inform people about how recommendations are generated and allow to edit them?

How might we...

...make sure people's data is well interpreted in order to build up accurate profile

Design Features
In context demonstration of how profiles are built

When an article has been recommended to someone based off their viewing and browsing habits, this is flagged and explained to them in-context. At a glance, they can see the tags that Bizfeed has associated to their profile based off their viewing history.

Buenos Aires - Bizfeed - Design Feature 1
Design Features
Instant editing of recommendation profiles

People can then remove tags that they either aren't interested in, overloaded with or that they feel may misrepresent them. The app also allow them to add tags to their recommendation profile themselves, making it even more accurate.

Buenos Aires - Bizfeed - Design Feature 2
Next steps

How might we build on Bizfeeds ideas to:

  • Use visualisation to explain how inferred data was gathered, with additional depth