September 6, 2017

Facebook memories: the research behind the products that connect you with your past

By: Artie Konrad

I recently opened up Facebook’s memory feature, On This Day, and savored the memory in front of me. I embraced the symphony of emotions that began to swell – stress, excitement, exhaustion, relief – the bittersweetness of farewells and new beginnings. On This Day reminded me about the moment I defended my dissertation and completed my Ph.D. I relive the experience of lying in bed when it was all over, feeling a vague sense of sadness that the 5 year journey had concluded, but elation that a new chapter as a researcher at Facebook was just beginning. This was an important moment in my life story, and On This Day provided the memory cue that helped me to experience it again.

Memories are complex, personal, and emotional. It is this very kind of complexity that I researched in grad school and have devoted my career to understanding. One area of interest, for example, is how positive experiences can be remembered more negatively over time (such as after a romantic breakup) and negative experiences can be remembered more positively (such as a new relationship that makes you feel better about an old breakup). These contamination and redemption sequences, respectively, as well as attitudinal factors can influence how one remembers things (McAdams et al., 2001; Konrad et al., 2016).

When On This Day was launched over 2 years ago, we came face-to-face with this complexity. What should Facebook’s role in reminiscence be in the first place? How do we approach Facebook memories with sensitivity and care? We need to be mindful that we aren’t just stewarding data, we’re stewarding autobiographical memories that form the stories of people’s lives. To help us answer these questions, I’ve conducted over a dozen quantitative and qualitative memory studies over the past two years. Research is the tool we use to help us traverse the space between heart and technology; connecting what people on Facebook desire, with what developers of our products create. This helps ensure we’re moving in a positive direction and building empathy on our team for people’s experiences.

On This Day has come a long way both in terms of quality, and in terms of how people have embraced memories on Facebook. I’d like to give you a behind-the-scenes peak at the research I’ve conducted. Additionally, I’ll summarize some of the feedback we got from people like you, that helped guide our efforts in developing two new Facebook memory products. These were designed to help people who might wish to experience their pasts in new ways.

The early days of Facebook reminiscence

The idea of a memory product on Facebook emerged to fill a need by helping people re-experience their posts without having to manually sift through their Timeline. For example, in 2014 we created the Lookback video and Say Thanks video, which told stories by combining memories, creative elements, music, and simple designs. Through research we learned that the story told in these videos, and the quality of the creative elements and execution, contributes greatly to the nostalgic impact they have. Contrast these early videos with more recent videos like the one you receive after your birthday or 2017’s Year in Review video, and you’ll see richer narratives, elaborate sets, and vibrant colors.

Lookback and Say Thanks videos (left). Personalized Birthday and 2017 Year in Review videos (right)

Defining our role

Shortly after testing the waters of nostalgia with Lookback and Say Thanks, we developed On This Day to help people revisit memories from a given day in their Facebook history.

On This Day

I invited people into our research labs at Facebook Headquarters and asked them what they thought our role should be in reminiscence. In summary, they told us “to provide reminders of fun, interesting, and important life moments.” So we set out to figure out what makes a memory fun, interesting, and important. To answer that I conducted three different styles of research:

  • First, I had people classify their memories into various themes such as achievement, vacation, food, and family. This is known as a qualitative study because a small sample of participants helped generate distinct categories of memories.
  • Second, I verified these distinct categories in a larger quantitative survey sent to people on Facebook who had recently seen a memory. I had them classify the memory into a theme (generated in our qualitative study), and then rate how much they enjoyed seeing that memory. I learned, for instance, that pictures of family are memory gold, especially kids because there is surprised delight in seeing how much children have grown over the years.
  • Lastly, I conducted a linguistic analysis of words used within anonymized memories to understand what types of content are more likely to be shared. Memories that had words like “miss” in them (i.e. “miss your face”) were more likely to be shared, whereas food-related words (i.e. “best taco ever”) were less likely to be shared because they were no longer relevant. What emerged was a framework which we call a Taxonomy of Memory Themes that not only helped us bucket memories into distinct categories, but began to inform our ranking algorithms that select which memories to resurface.

Taxonomy of Memory Themes

Overall, when I am conducting interviews with people, I make sure the memory team is closely engaged and observing these sessions to foster their empathy and understanding. It’s one thing to provide feedback to a designer about the experience they are building, and an entirely different thing for them to hear straight from the people that will eventually use the experience on Facebook.

Respecting the boundary lines

Now that we’d done the work to better understand what our role IS in resurfacing fun, interesting, and important memories, we needed a better understanding of what our role ISN’T. In other words, where’s the boundary and how do we avoid overstepping? For this, I again brought people into the lab and conducted a deep-dive to identify a handful of boundaries, and each identified boundary became a separate branch with its own dedicated research. Here are two examples:

  • NegativeMemories: There are certain types of posts that people would prefer not to see surfaced again as memories. For example, people might post about violence, accidents or failed ventures, which can be rather unpleasant, particularly to re-experience as memories years later. We’ve been working on ways to better identify these types of memories and filter them out, so that people see more enjoyable content instead. However, even with the considerable progress we’ve made, we know there will always be the possibility that positive memories that have become negative – or contamination sequences – slip through. For example, a post about a first day at a new job that has lots of positive reactions and comments may have become negative over time if the person has left that job. Thus, we learned we also needed to give people more control over the memories they see.
  • Lack of Control: In our research, people said we would be crossing a boundary if we didn’t provide them with control and options for how their Facebook memories appear. There needs to be options to help people adjust On This Day so they can remove contamination sequences and other memories they might prefer not to see. To try and meet that need, we built a set of preferences designed to empower people with the ability to let us know the specific people and dates they prefer to filter from their Facebook memories:

    Memory Preferences

    We have since conducted research on these controls to help us make them more easily discoverable, accessible, and useful to people and will be rolling out these changes in the coming months.

    New directions

    As a researcher at Facebook, my job isn’t just to present data to product managers, designers, engineers and other developers once my studies have concluded. My job is also to try and help bring all these stakeholders along for the entire research voyage, so we can listen to feedback as a team. That’s no less true with the two new experiences I’m excited to introduce today, which were inspired by our research and conversations with real people.

    First, while older memories contain an element of surprise and nostalgia, people expressed interest in revisiting more recent memories to help them enhance and prolong their enjoyment. I’ve seen in my experiments conducted before working at Facebook that recent memories may sometimes have a positive influence on mood and well-being through various mechanisms like savoring (i.e. Isaacs, Konrad et al., 2013). We built on this idea and created a new memories experience that packages your recent memories in a delightful way for you to enjoy and share. To arrive at this concept, we conducted three rounds of research to better understand the appeal and iterate toward a useful and enjoyable experience. We found that recapping a recent time period (like summer) gives people an easy way to summarize a chapter in their lives for themselves and others. This was a natural extension of what many people were already attempting to do on their own – like creating albums of their best summer moments to look back on and keep their friends current on their lives.

    Monthly (left) or Seasonal (right) Memory Recap Story

    Aside from reminiscing on recent Facebook posts, we also learned that there are Facebook activities that are meaningful to people. Over the years we’ve heard numerous inspirational stories from research participants about how Facebook helped them reconnect with long lost friends, and create new friendships that might not have been possible without the platform. These connections have meaning to people, and it’s that meaning which inspired an experience to celebrate the actions that connect friends, family, and their communities on Facebook. Initially, this experience was meant to encourage reminiscence on significant Facebook activity such as the number of friendships you made this year, or likes received on your posts. However, when we tested these in research, people felt that the numbers alone lacked some of the heart and soul that makes the experience meaningful. We took this feedback seriously and evolved the experience to be less about the statistics, and more about the people and memories behind the numbers. For example, if your posts received 100 likes this year, the product surfaces the actual posts that received the most likes, celebrating the moments themselves that bring value to the number.

    Friends Made Message (left) and Likes Received Message (right)

    At Facebook we like to infuse ideas with multi-stage research across a variety of creative methodologies, build empathy for people on our platform by engaging with the research collaboratively, and listen to feedback from the people who know themselves the best. We’ll continue to think deeply about these experiences and our sensible stewardship of memories, and we hope you’ll continue to find joy in remembering the many life moments you share on Facebook.

    Artie Konrad is a User Experience Researcher at Facebook interested in the intersection of memory, emotion, and technology. He specializes in Technology-Mediated Reflection (TMR) which is the use of technology to support reminiscence on digital memories.