Growth team

Product design

Prototyping

My first project at Facebook was to refine the “Search for friends” step in new user sign-up, a critical step in user retention funnel.

Data on user retention showed that if new users couldn’t find friends once they’d singed-up, they would leave the service, and rarely return.

In 2014, Facebook was focused on growth in markets where mobile-web was the primary way the service was being accessed. Hence, this was where the team felt we could have the most impact.

This was the current mobile-web experience for adding friends during sign-up:

• 350,000 people visited this step daily

• Only 28% will make at least one search

• 63% of those who search click on see more in the results

• People who click on see more will do so an average of 6 times

• Of those who do search, they will make an average of 3.4 searches

• 88% of searches result in a friend request

We performed user research and identified user problems which helped explain the data:

• People don’t understand why they need to add friends on facebook

• It’s tough to find people they are looking for in search results and have to scroll and see more until in order to do so

Given the above, I isolated and ordered objectives (based on impact) for the experience:

1. Increase the number of people who will make an initial search

2. Improve the quality of search results

3. Increase the number of friend requests sent

Below is an overview of the new experience I designed and prototype:

User Problem 1: Why should I add friends on Facebook?

I explored multiple solutions, including educating users on graph search, which would help improve the quality of results (prompting users to search for Brian Smith Indiana State vs Brian Smith). We also employed a facepile background to give users a sense of community.

A prototype below:

This was problematic for 2 reasons:

• Graph search is not intuitive to users, especially new ones, and can overwhelm the user which results in them skipping the step entirely

• Refining search results is down-funnel from making an initial search (our top line goal) so we should prioritize the up-funnel experience of making an initial a search

In order to engage people to make an initial search we wanted to give them a sense of community and clear value prop to do so:

User Problem 2: It’s tough to find who I’m looking for in search results

A significant portion (88%) of people who search make a request, so we didn’t want to add friction to this step.

To accomplish this we added a contextual graph search suggestion after a search returned too many results. Upon tap - the contextual aid would focus the user back into search and suggest adding key words to narrow results.

Objective: Increase number of friend requests sent during sign-up

While not a user problem, this was a business goal. When a user has at least 10 friends, they will find value and continue to use the service. Our head of growth, Alex Schultz, called this “the magic moment.”

In service of this objective, we wanted to show a people you may know step before search for friends. We would only do so if we felt we could show you good recommendations.

I designed this step, adding in unique new component - a row of blank state friend cells which appeared after the user had sent one friend request. With this design, there would be a subtle cue to add five friends after adding one.

We were excited about the impact of this work on growth. After this initial experiment, we would follow up with work on iOS and android clients. My internship ended before I could see experiment results. Shout out to Scott Savarie and Vlad Milosevic for all the great feedback along the way, and to Joey Flynn for the quartz composer lessons.