Skin Tone Filters

Motivation

“I use social media and am heavily influenced by it. I want to feel relevant. I want to feel beautiful. However, the beauty standards portrayed by Pinterest have no people of colour. If I ever want to feel like I am a beautiful contribution, I cannot look at your app.” - Anonymous Pinner

More than 70% of people use Pinterest to discover and save beauty ideas they want to try. However, a common user complaint stems from a lack of diversity in our beauty results, which is easily illustrated by the search for “dreads.”

The results are less than ideal. A user who wants to see darker-skinned women in their results needs to prepend “black women” to their text query, a process that is time-consuming, non-discoverable, and most importantly, non-inclusive. Furthermore, even after a manual change to the query, the results are often either not relevant to the skin tone specified or the original text query.

To help users better find what they want we specifically enabled users to seamlessly filter beauty results based on a user inputted selection of skin tone range.

Goals

1. Build a more inclusive search experience
2. Help Pinners more seamlessly find relevant beauty ideas in search

Success Criteria

• ↑ search success rate
• ↓ number of re-queries
• ↓ user complaints
• [Guardrail] maintain the relevance of search results
• [Guardrail] filter engagement and retention rates (baselined off of other filter usage)

Results

We saw amazing metrics wins with respect to our success criteria. We also saw an increase in overall search relevancy, overwhelming positive reactions from qualitative research and in-app surveys, and a huge marketing moment win for Pinterest with dozens of articles published on the feature launch. But most importantly, we allowed users to find what they wanted, as illustrated in the new search for dreads with the darker skin tone selected.

Learnings

1. There is a large opportunity to improve search by offering more customized filters. This feature was extremely effective in helping users narrow and find what they wanted and only represents a small portion of facets used to refine searches.
2. User research early and often. We learned and iterated before we even built the feature by exposing internal employees and a handful of users to the idea and mockups. In general, it diversifies inputs to the project and can either challenge or support intuition quickly and with relatively low effort.
3. Motivation is crucial. I worked with a team of extremely passionate people on this project and it was by far one of the most rewarding and smooth projects I have led. While the motivation did not necessarily come from me, the takeaway here is just how crucial it is to correctly motivate and evangelize a product.