Scaling signals: using qualitative research to influence algorithms

Nate joined us at UX New Zealand 2019. Find out more here: http://www.uxnewzealand.com/speakers/nate-foulds/

At Instagram, they use research to find signals in the wild - about people’s backgrounds, behaviours, likes and dislikes - that can be incorporated into the algorithms they use to deliver content.

A few years ago, Instagram launched its algorithmically sorted Feed to the outrage of many users who preferred the chronological version. In data, however, they saw that instead of missing out on content, the ranked Feed actually increased the percent of posts people consumed from their friends and family. In both Feed and Stories, two of the most highly consumed surfaces in the world, it’s crucial that Instagram deliver the right content to the right people at the right time.

In this talk you’ll learn about how qualitative research is leveraged to improve the strength of the algorithms that power Feed and Stories. For instance – how can they predict which friends’ content is most appealing to you? Instagram use ethnography, interviews, diary studies and more to gain insight into the signals that matter. Using these signals, researchers work with engineers to boost levers in the algorithm and test their hypotheses. By the end of this talk you should have ideas for how you can uncover your own signals and measure their impact in the wild.

Opening animation by Dave Clark Design. http://www.daveclark.co.nz/