Two teams, one thread: How Figma pairs data with user interviews

Figma’s data science and user research teams found that their best insights come from a process they liken to weaving. Just as weft yarns cross over warp yarns to create cloth, quantitative and qualitative findings need to intersect to form a complete picture. Data scientist Caitlin Hudon and researcher Jennifer Sanders spent several weeks combining their disciplines to understand how notifications work—and where they fall short.

Notifications in Figma reach users through email, Slack, mobile devices, the file browser bell, system tray bell, or desktop bell. They cover a range of events, including comments on files you own or contribute to, replies to comment threads, reactions to your comments, invitations to files/teams/projects, edit invitations, and at-mentions.

Notifications flow through a funnel: users must have them enabled, receive them, view them, and finally interact with them. The activity team—part of the teamwork pillar focused on helping teams stay connected—didn’t know where in the funnel to concentrate their efforts. Hudon dug into the funnel from a data perspective while Sanders ran research on the same topic. Their overlap revealed a chance to build a story stronger than either discipline could tell alone.

“Quantitative data tells us the what,” says Hudon. “We can dig into numbers and see user behavior at scale, but those insights are more helpful when we start to understand why users are doing those things. The best way to figure that out is to actually talk to them.” Sanders adds that why is tricky because “people aren’t always good at explaining it.”

Structuring the collaboration

The pair started with a team-wide brainstorm in FigJam, presenting current notification metrics and setting quarterly goals. The team wrote down strategic questions, ideas for alert types, and data gaps. Afterwards, Hudon and Sanders split open questions by which discipline was best fit to answer each.

With individual research projects underway, they relied on steady communication rituals:

  • An ongoing Slack channel for sharing discoveries and asking questions
  • 30-minute work sessions to trigger notifications in their own Figma files and hash out developments
  • 60-minute work sessions for deeper interpretation of findings and joint share-outs

Questions begot more questions. "As we understood our domains better specifically around notifications, we started to go one step deeper," says Hudon, recalling wondering whether an observed behavior was the exception or the rule. Sanders echoes that it was iterative: "Our data kept pushing us to ask more questions."

What the numbers and interviews revealed

Their iterative layering exposed discrepancies between the two views of user behavior. The numbers showed few users engaging with email notifications, yet nearly all of Sanders’s interviewees reported relying heavily on email to stay on top of work. That mismatch was the first clue. Hudon soon found the reason users weren’t interacting with email notifications: most weren’t receiving any at all.

Sanders’s unexpected findings led Hudon to examine team-account users as a separate segment. The cut revealed a very different population: designers on team accounts were opening 80% of the comment notifications in their inboxes. This insight became a strategic recommendation: notification experiments should center on users with team accounts, since they collaborate more and their behavior better reflects how notifications should work.

Finding the bell problem

One question drove early investigation: do users notice the notification bell in Figma’s file browser? Hudon could see which users clicked it and how usage compared to email or Slack channels, but not why the numbers were low. Sanders’s research revealed the bottleneck wasn’t awareness but workflow—users checked their email inboxes first thing in the morning across all projects, and kept Figma files open in tabs to toggle between them. The bell, hidden in the top right corner, was "out of sight and out of mind."

This shifted the half’s focus toward making the bell more noticeable and improving notification content and reach. The pair recommended new alert types to bring users back to files, inspired partly by Hudon’s own experience: she contributed to a FigJam for a joint presentation but received no comment notifications afterward because she wasn’t the file’s owner. The team saw the missed opportunity.

From recommendation to release

Based on these findings, the activity team ran an experiment to send comment notifications to all file editors—not just owners. The feature has since shipped, allowing collaborators to follow activity on files they’re invested in.

The cross-functional insights helped communicate strategy and shape the roadmap, with concrete examples from both data and customers. "I can throw those numbers, but that’s not nearly as clear as shortcutting that with one insight from a user saying, 'The notification bell doesn’t fit into the way I’m using the product,'" says Hudon. Sanders points to the data side: "The punch doesn’t always come from research; sometimes the punch comes from data. For example, '71% of users don’t receive any notifications.'"

Both emphasize having a dedicated partner who can represent your data in meetings you miss, and who brings the full ecosystem context to evaluating insights. Combining qual and quant isn't revolutionary, Sanders admits, but the meticulous synthesis often gets skipped in the interest of speed. "Investing in the synthesis as much as we did, as quickly as we did, to ensure it would have impact—that’s what differentiates the results."