We recently published the world’s largest benchmarking study of Workplace by Facebook networks, covering 68 organisations and more than 630,000 people. In this report, for the first time, we have extended our analytics to the Workplace group level (Teams & Projects), Communities and Q&A Forums (Discussion) and Announce and Social (non-work).
One of the immediate findings from the research into our sample of 1,360 teams is that digital teams are way too large to enjoy the benefits of being a truly high productive team. At an average size of 296 members, there is little chance of achieving the level of cohesion and productivity that more appropriately sized non-digital teams do.
We believe a fundamental shift is required in how digital teams are formed on Workplace. If teams are to truly perform online, they should limit their size to less than 10 members. It will be a change in practices, more so than technology, that will be required to achieve the levels of productivity anticipated from dispersed digital teams.
Best practice Teams
J Richard Hackman is widely acknowledged for his 40+ years of research into teams. It is his rule of ‘no double digits’ for the most appropriate team size that we have locked onto for our use. In essence, Hackman says no more than 10 members for a team, but 4-6 is ideal. Amazon Founder Jeff Bezos’s ‘Two-Pizza’ rule more colourfully suggests that a team should be no bigger than can be fed by two pizzas. The research by Sandy Pentland and his team at the MIT Human Dynamics Lab broke new ground in data driven analysis of team performance. Pentland used custom-designed social tags to capture human interactions within teams at a forensic level of detail. While the mining of interaction data from Workplace might be a somewhat poor substitute for MIT’s social tags, the breadth of monitoring available to us is substantially greater. In his book “Social Physics”, Pentland reports on many social tagging studies, identifying those social signals that are most aligned with increased productivity or creativity.
Like Pentland, the team at Google directed its massive analytics capabilities to some 180 of its own teams, with the somewhat surprising finding that the biggest predictor of team performance was ‘psychological safety’ i.e. the freedom to be open, honest and authentic.
The concept of Psychological Safety is treated in some detail by Harvard’s Amy Edmondson in her recent book “The Fearless Organization”. Edmondson collates decades of academic and industry research identifying the influence of Psychological Safety on team productivity and/or creativity.
Collectively, the academic research provides us with a rich suite of limited sized studies that link particular team attributes to higher levels of productivity and/or creativity. Only Google’s Aristotle project approaches this study in terms of breadth; though addresses only a single high-tech organisation, not particularly representative of the general population of organisations we represent here.
Unlike the above studies, SWOOP has no access to individual team business performance information like increased productivity or creativity. We have instead looked to align our own SWOOP measures, where possible, with characteristics identified from the research. We then apply these aligned measures to our collection of 1,360 teams collected across the breadth of our 68 benchmarking partners. For those teams that prove strong performers with our selected metrics, we infer are high performing teams.
Some measure of validation is being attempted through direct connection to these teams, though this is an on-going task. How teams are distributed across the team performance spectrum will provide us with a ranked list of Key Performance indicators that teams can assess their own predictive success factors.
We draw from the research findings to provide the following aligned team performance criteria:
| Team Characteristic | SWOOP Metrics | Commentary |
|---|---|---|
| Has no more than 10 members (Hackman, Bezos) |
• Team Membership • %Active Users (Participation) |
It is common for digital teams to be bigger than they need to be. It takes some discipline to limit a team to only those that can actively contribute. |
| Bound by a sense of common purpose. (McChrystal, Denning, Hackman) |
• Two-way Relationships • Response rate • Persona Distribution (Engager Proportion) |
A common purpose can galvanise teams into a higher level of co-operation. Two-way relationships are our proxy for enterprise ‘trust’. Fast response rates and a high proportion of Engagers infers interactions fired by a common purpose. |
| Everyone on the team talks and listens in roughly equal measure, keeping contributions short and sweet. (Pentland) |
• %Active Users • Activity/User • Influencer Risk • (Reply) Response Rate |
The active users identify team member participation levels. A low influencer risk score infers equality of contributions. Fast response to posts infers short sharp interactions. |
| Members connect directly with one another—not just with the team leader. (Pentland, Denning) |
• Influencer Risk • Persona Distribution |
The influencer widget shows changing influencers i.e. not always the leader. A spread of members listed with most engaging posts infers broad-based interactions. A mix of Engagers, Responders and Catalysts with low proportions of Broadcasters and Observers also infer broad-based interactions. |
| Members carry on back-channel or side conversations within the team. (Pentland) |
• Activity levels • Mention Index |
Team activity identifies the side conversations happening within the team. Mentions identify members being brought into (side) conversations. |
| Members periodically break, go exploring outside the team, and bring information back. (Pentland, Google, Denning, Hackman) |
• Diversity Index • Curiosity Index |
Diversity measures the degree to which team members are also members of other groups. The curiosity index infers a level of exploration. |
| Psychological Safety (Google/Edmonson, Hackman) |
• %Active • %Mention Index • Two-way relationships • Diversity Index • Curiosity Index |
We infer a psychologically safe team exhibits strong cohesion (Two-way relationships) that consistently ‘tag’ others into conversations; high participation by all members; members with diverse experiences and a high level of curiosity. |
What do High Performing Teams do Better?
If we drill further into the data, we can find what factors appear to separate those higher performing teams from the rest. We do this by calculating a standardised variance for each indicator and then ranking them. The higher the variance, the bigger the gap between the best and worst; and therefore the greatest scope for improvement.
The Replies/Post is a responsiveness indicator. Therefore, the greatest difference between high performing and low performing teams is their responsiveness.
The next highest is the Threads/User, which is a measure of the breadth and inclusiveness of discussion. High performing teams just talk to each other more.
The third indicator, and arguably the most problematic is the team size. The average sized Workplace ‘team’ is 296 members. When compared with the recommended maximum team size of 10 members, there is clearly an issue with not just Workplace teams, but all digital teams.
We have written previously about the tendency for digital teams to grow, simply because it is easy to do so. We would never consider forming an off-line team with 300 members. Yet we have an expectation that online digital teams will perform as well, if not better than regular off-line teams. As it stands, the vast number of Workplace teams are operating simply as information sharing spaces for anyone who may be interested. Which we think is a pity. Indeed, the data shows that performance does drop off as team size grows:
We believe a fundamental shift is required in how digital teams are formed on Workplace. If teams are to truly perform online, their size should respect the science and limit their size to less than 10 members.
It is not possible to facilitate the psychological safety required to build a high performing team when the membership grows too large. We have observed that organisations that have natural teams limited to less than 10 members e.g. airline crews, bank branches, fast food shops etc., typically align their ESN groups and demonstrate high cohesion scores.
If your team has more than 10 members, think about spinning out a core team of less than 10, and ideally between 4-8 members, and re-labelling the existing team a Q&A/Community group. In this way the smaller team can freely interact in a psychologically safe environment, and then choosing the content and timing of what they share with the broader stakeholder community.
How are your Teams Connected?
US General Stanley McChrystal coined the term “Team of Teams” to describe how he facilitated the re-engineering of the military forces in Afghanistan to meet the need for extreme responsiveness and adaptiveness. Organisational change guru, John Kotter, in his book XRL8 paints a similar picture of organisations needing to balance structure and hierarchy with internal networks designed for responsiveness and adaptability.
This is exactly the challenge that users of Workplace are looking to overcome. Workplace provides the platform that a ‘Team of Teams’ organisational architecture can be established. While Workplace can provide analytics on individual team activity, there are few - if any - tools that can provide a bigger picture of how teams are actually connected.
In the graphic below we have taken SWOOP benchmarking data from one organisation, extracted the ‘Teams’ (less than 20 members) and then created a ‘Team of Teams’ network representation based on overlapping memberships. We found the proportion of Teams existing in a single business unit and the teams comprising members from multiple business units was split 50/50. We also found a similar proportion of teams that were totally disconnected.
Even these simple measures can provide you with a picture of how well your organisation is balancing the discipline the hierarchy provides with the responsiveness and adaptability a Team of Teams structure provides.
The architecture below suggests how Workplace teams can get to the optimal size.
Connection between teams should be via common members who have shared roles, e.g. program managers, product owners, QA officers etc. Communication with the broader stakeholder groups should be via stakeholder forums or communities.
Workplace Team Targets
To assist Workplace team leaders, and especially SWOOP clients, we have developed a set of recommended targets for selected team key performance indicators. Each of these indicators can be found under the Groups Tab on the SWOOP dashboard (Select a most recent six-month period). Other than Team Size, the targets have been set to the level achieved by the top 10% of current teams assessed.
SWOOP Metric TABLE NEEDED HERE
* Influencer Risk target is a maximum risk score.
Based on the data we have analysed, only 5% of Workplace teams have less than 10 active members. We anticipate that as you reduce the size of your teams to the optimal level, the stated targets will become easier to achieve.