Facebook Reels

I was brought into the Facebook Creation team to lead the Reels product from 0 to launch. TikTok had popularized the short-form video (SFV) art form, which had recently been reimagined on Instagram. Facebook, whose audience had unique characteristics and behaviors on the Facebook platform, wanted to develop a unique approach to the SFV market.

This page covers my contributions to the development of Facebook Reels, from product placement to use-case coding to benchmarking, this project was a soup-to-nuts collaboration across various Meta domains.

Role & Team

I led research for the Creation team, which impacts how users create the content that appears on Facebook. I collaborated closely with designers and PMs across various teams within the Creation ecosystem, such as Home, Identity, and Consumption, as well as Instagram researchers. This included content designers, prototype designers, PMs, data scientists, and various stakeholders and leaders within the product space.

My core research team was comprised of 18 researchers, all with various foci in the creation space on Facebook. I led two other Creation researchers and mentored a contractor during my work on Reels. Many of them were terrific researchers, and we shared often to bolster and rebuff findings. I really loved the research culture and dynamic at Facebook - it was rigorous, energetic, and open.

Reels Research

My manager liked to keep researchers’ plates full, and I typically balanced 1-3 iterative projects and 1-2 foundational projects at any given time. The result was a smorgasbord of research across various stakeholders under the umbrella of Reels:

  • After onboarding, I got to work understanding Reels with a baseline study. This kicked off my influence and direction for the next research phases.

  • I got to work with some great researchers, designers, PMs, and data scientists on various Reels efforts, including a project that impacted the top of the Facebook homepage that Mark Zuckerberg was interested in.

  • I dove deep into Reels as a creation format, and learned how limitations can be key to creative expression, as well as defining user’s mental models of product boundaries.

1 - Reels Baseline

I began my Reels research by baselining user expectations and impressions when creating a Reel (which already existed in India at the time as a model testing population). Working with a more experienced Facebook researcher on test account setup and screeners, I completed a groundwork study to address stakeholder questions around camera tools, creation mechanisms like entry points, and sharing impetus. I ran a study using UserTesting.com with ~12 users in India that had previous experience watching Reels. The results impacted entry points, including consumption and creation, camera orientation and usage of pre-existing content, and funnel issues related to sharing and identity, as well as drafts. This led to an overall insight that Reels lacked a distinct product understanding within the product ecosystem, causing confusion with other content formats and rationales for creation.

2 - Reels on Facebook Homepage

Part of the issue at Facebook with Reels in particular was lack of awareness of a consumption entry point. At the time, Reels intermittently showed up on some people’s feeds, but others had entry points in the bottom row, overflow menu, or other locations. One destination of interest was the top of the Facebook Homepage. Based on my baseline, I led a follow-up effort with the Home and Consumption teams to study of a new widget at the top of the Facebook homepage (then known as the most expensive piece of digital real-estate). I ran a study with 24 users using remote 1:1 interviews to answer questions around where users thought Reels, Stories, and other shared media content should live on the Facebook app, and the impact of consolidation into a top of feed unit, as well as interactions and impressions for creation and consumption, with guidance provided by alternatives. A resulting widget launched at the top of Facebook Feed in an attempt to disambiguate content types, but had moderate lift compared to previous mechanisms.

3 - Media Sampling

To better understand how the Facebook Reels product was evolving, I spearheaded a media sampling effort to catalogue types of content by various criteria, such as like or view count, subject matter, and user type. This helped our data science team and lead Reels PM to understand how our product was being used in the wild, and could help identify meaningful content types, camera orientation vs upload of content, and other factors we could leverage to increase DAU/MAU, leading to product growth. I worked with a team of 6 contractors to provide analysis guidelines for the samples viewed, which I pulled with the collaboration of a data scientist to get representative clusters within content types. The results I analyzed using a Shiny App in R and outputted a document detailing the findings to the Reels PM Lead.

4 - Quant Consumption Question

A continued hypothesis the team had was that watching SFV led to greater creation via inspiration. Thus, if we could get more people to watch Reels, we would see greater content creation. I devised a strategy to analyze Reels consumption and creation trends to test this hypothesis. Working with our data science team, I was able to gather a CSV of time-stamped flows, which I parsed and filtered to showcase users that 1) likely clicked to watch reels intentionally, 2) likely clicked create, and 3) actually completed the creation flow. The data here showed a statistically significant inverse correlation in dwell time to creation success. Moreover, it showed that most users created within the first few Reels consumed. While this does not show that consumption stifles creation, it did show that it may hinder user’s willingness to try to create during consumption. This led to several new hypotheses which were useful to the Creation team: users likely already have content in mind to create before watching Reels, and this may be why we were finding more uploaded content rather than content created and edited within the Reels Creation editor.

5 - The Creation Model Framework

Based on my legacy of work on Reels in various contexts (including what’s not shown here: Stories to Reels), I undertook a much larger initiative, partnering closely with a design lead, in which we ran a design sprint and tested concept directions for new creation models. Our goal was to define new ways for users to create content on Facebook, and to further disambiguate Reels from other Facebook product offerings. While I led portions of the design sprint, such as interviews with key stakeholders to understand creation history and learnings for the team, my main contribution was a post-sprint research effort to study the concepts that the team had outputted. The sprint resulted in many concepts, so I worked with the designer to output the visions into 3 variations to test limits, comfort, outputs, sharability, and other factors.

6 - Creation Funnel

Getting more in the product weeds, I also partnered with various Creation designers to work on components of the Reels creation funnel, like building based on a template, finding popular trends, or converting past Stories to Reels. Another noteworthy investment was a large-scale effort I co-led with a researcher from the Instagram Reels team to benchmark the Reels Creation process compared to other top competitive platforms, like CapCut. Our goal was to understand the variety of features these apps had across a variety of Reels scenarios and benchmark them. To do this we partnered with MeasuringU and provided robust direction as to the audience (screener), content creation types (with examples, largely borrowed from the media sampling activities), and metrics of interest. The output helped inform tools and features to focus on, as well as what to not focus on - always an important distinction. We incorporated these results into further studies of our Creation funnel for evaluative testing.