We recently completed what we believe is the largest study of digital team collaboration performance ever undertaken. We analyzed 1,360 digital teams from nearly 70 organizations from a full breadth of industries and geographies. The digital interactions of over 400,000 team members and nearly 2 million interactions over an extended 6-month period were analyzed. Our sample was drawn from users of the collaboration platform Workplace by Facebook, where groups have been self-identified as a team.
Indicators of Successful Teams
We wanted to understand how digital teams were performing against the benchmarks established through decades of academic research on teams. The most basic measure was optimal team size. Harvard "teams research" guru Richard Hackman suggested this is less than 10 and optimally between four and six. Amazon CEO Jeff Bezos's “two-pizza” rule reinforced this pragmatically, likening optimal team size with the size of their food order.
MIT Professor Sandy Pentland coined the term “Reality Mining” to describe his use of digital social tags to monitor social interactions within teams. Pentland identified performance factors like reciprocity of interactions, exhaustive peer to peer interactions, and a diversity of experience where ideas are sourced from beyond the team.
Finally, Google’s much publicized Aristotle project analyzed 180 teams to find that most of the anticipated key indicators — team composition, skills, friendships, etc. — could not consistently predict team performance. The key predictor they found was “Psychological Safety,” which basically means team members feel safe to speak freely, without negative impacts in their teams. Harvard professor Amy Edmondson provides decades of evidence from research around psychological safety in her recent book "The Fearless Organization."
Our Study Methodology
For our study, we wanted to understand the extent to which digital teams were demonstrating the characteristics these academic studies had identified as predictive of highly productive teams. We selected online measures that approximated the factors identified in these studies. Online team measures included reciprocal relationships, diversity, curiosity, participation, tagging, response rates and my firm, SWOOP's, behavioral personas. We then rated each team and ranked them by performance.
Related Article: Will Customers Buy Into Your Seller Persona?
Team Size and Psychological Safety Are Key
The most fundamental insight is around team size. The average team size was 296, with one team having over 1,000 members. From this we can see that staff are not equating a digital team with a non-digital team. Who would purposefully form a non-digital team of nearly 300 members? It fails the two-pizza rule by a county mile!
We can’t point to Workplace as the culprit. A 2012 IBM study of their online teams found the average size of the 73 teams surveyed was 416 members. While we don’t have comparable statistics from "team" tools like Slack and Microsoft Teams, we know there are no technical limits to creating a team of more than 1,000 members. We see regular reports on issues with large Slack teams in terms of the "digital noise" created for the majority of team members, to the extent that Slack is providing advice on how to manage large teams.
It appears that online digital teams are not being formed in the same way, or for the same purpose as non-digital teams. More likely they are being formed as an information sharing adjunct to non-digital teams, which we think is a pity, and a gross underuse of the potential digital team tools can provide to teams. The onus is on those creating digital teams to consider their productivity and the research advice on highly productive teams. Keep the numbers below 10 and aim to create the psychological safety that high performing teams require.
Related Article: Collaboration Is Key to Distributed Team Success
We followed the guidance of Harvard's Edmondson and the experience from Google’s Aristotle project to devise a set of measures that might infer a psychologically safe team environment:
Psychological Safety Indicator
As per the research; < 10 members is best.
% Active Members
Ideally all team members should be active online.
% Two-way Relationships
Measures the reciprocity between members. Ideally 100 percent.
% Response Rate
Measures the rate that posts get responses. Ideally should be 100 percent.
Overall Activity Levels
We would expect a high number of interactions every day.
% Engager Persona
The SWOOP engager persona reflects people who help others connect. A high percentage is best.
Measures the proportion of posts framed as questions. The more questions, the more likely the environment is psychologically safe.
The diversity measure indicates the degree to which team members have been active in other groups; signally a diversity of experience.
Measure the percentage of messages that include a mention of another team member; essentially ‘tagging’ them into the conversation.
We believe that this collective suite of measures, if maximized, would provide a psychologically safe, and therefore highly productive team environment.
For simplicity sake we rated each factor equally and then assessed the 1,360 teams against a composite index of these measures. The following distribution identifies the relative performances:
While no team was perfect (100 percent performance index), we found the top performers (roughly 10 percent of the sample) were significantly better than the remainder. The average team size for this top 10 percent was 23 members with the top 10 averaging below the recommended 10 members. So size really does make a difference.
Related Article: The Manager's Guide to Transparency in the Workplace
What Can You Do?
If you have a digital team of more than 10 members:
- Form a new team of less than 10 from the core participants.
- Use measures as indicated above to create and monitor the psychological safety of your new team environment.
- Retain the former team as a stakeholder community for the team, where you can post updates and have more open discussions.