SambaNova was looking to help employees find ways to stay connected as they grew. Relying on chance encounters was not an option, especially given the company's hybrid work model, and employees were not finding random weekly 1:1s to be valuable.

SambaNova's People Ops team brought Thred on to help with two main goals: identify and support potentially at-risk, isolated employees.
break down silos, create paths for sharing knowledge, foster a sense of belonging.

We began with a classic organizational network analysis (ONA), looking at who knows whom—and how well. From our initial visualization, we quickly found who was isolated and in need of extra support.


From there, we looked at individual folks' networks and began segmenting into network personas (aka roles or archetypes). We started with 3 simple ones: Connector, Bridge, and Influencer.Alone, those personas are pretty interesting, but the really compelling stuff shows up when you layer them on top of one another. On this project, we spent a lot of time thinking about the groups that fell into Bridge or Influencer roles but not into Connector roles. These "hidden" personas are particularly hard to find without network analysis.The insights from the network analysis became inputs for the actual connections part of the project. For example, we realized that one of our Hidden Influencers - we'll call her Jana - was planning to retire in a couple months. We know that if a Hidden Influencer leaves, their main connections become turnover risks. So, we prioritized connecting Jana's main connections with Super Connectors who could help pull them into the center of the network and fill the space that Jana would soon leave.Now the fun part. We facilitated roughly 4 rounds of connections over 6 weeks for 200 SambaNova employees. Along the way, we gathered feedback about what worked well and what could have been better (more on that in the Results section).

Thred's approach to connectivity is driven first and foremost by shared interests. Our matching algorithm ensured that every single pair had at least one deep, specific interest in common (think: backcountry skiing, jazz, roguelike video games, etc.).To keep things simple, the actual 1:1s took place over Google Meet. The invites were customized based on shared interests, and we tried out a different format every week. Each week, participants let us know how things went. We were thrilled with the results that came back.

The number one takeaway was that after the program, half of participants expected to become friends with their match—and would not have met them without Thred. It is worth mentioning that while the majority of participants provided feedback, we, of course, didn't have 100% participation. Because of that, we do expect some response bias to be baked into these numbers, as those who feel strongly (positively or negatively) are more likely to respond.Value derived from: expected decrease in turnover, decrease in absenteeism, increase in productivityCost derived from: program cost and the cost of employees' time to participate Employee wellbeing and culture benefits not quantifiedWe used the match quality scores above to help determine expected ROI from the program. The general logic behind the calculations looks something like this:All said and done, the Thred program had a huge impact on the SambaNova employee network, especially given the relatively small time requirement from participants. The before and after of the employee network speaks for itself: