As we started introducing SWOOP to Workplace by Facebook clients, the big difference we noticed between Workplace customers and customers using other Enterprise Social Networking (ESN) platforms was the proportion of “Open” to “Closed” groups. Our 2017 benchmarking report of 57 organisations worldwide identified an average 74% open groups, with four organisations being 100% open.
We have always pushed to keep the number of open groups as high as possible, while acknowledging there are always good reasons to have some groups closed. Isn’t that why organisations are adopting an ESN in the first place, to enhance transparency and open sharing?
It was time to investigate why things were different on Workplace. We soon learned the answer. There was a fear your innocuous rant about the latest management initiative, made in the apparent safety of your close colleagues’ group, could end up on the CEO’s news feed, as it could on Facebook! Affectionately referred to as ‘group leakage’, if left unattended, could easily kill off your Workplace adoption program before it even started. Workplace therefore took the pragmatic step of advising their clients to ‘close’ groups, only leaving groups open for public announcement-type activities.
Perhaps the safety and privacy of closed groups really is where people get work done? I recall a conversation I had with a former organisation CEO, who was the sponsor of our global Knowledge Management program at the time. He was an engineer by occupation, having a long experience with large scale industrial projects. While passionate about knowledge sharing, he did make the point once to us that: “Silos aren’t bad you know; that’s how we get work done. We just need to drill some holes in the sides!” It was a good lesson in appreciating the balance that needs to be had when faced with a dilemma like Opening or Closing off a group.
While we had our reservations, it behoved us to run the numbers to see what the ‘real’ situation was. Just what is the impact of having a large number of closed groups? It was time to look for evidence in the analytics, as to what the impact might be.
Show me the numbers!
Armed with a Workplace data set from a SWOOP client with some 6,000 active participants and approaching 1 million interactions over a 6-month period, we selected groups with more than 500 activities in that period (141 groups) and assessed them on our Organisational Performance framework. Organisational performance was judged as the combination of a group’s diversity (based on the diversity of its members, measured by their breadth of activity across multiple groups) and its cohesion (measured as the average number of reciprocated relationships held within the group). The focus on diversity and cohesion is driven by the broad agreement among researchers, that groups where members interact well together, while also having exposure to goings on outside the group; deliver better business performance. Here is what we found:
Networking performance is maximised when both diversity and cohesion are maximised (top right of the chart). We can see here that several closed and highly active groups are leading the way. The majority of open groups are clustered in the top left, being the quadrant where groups are highly diverse, but not that cohesive. An open Workplace ‘Announce’ group would fit this profile. It is worth pointing out that most groups performed well in terms of diversity; meaning that the group members were active in multiple groups. Groups in the lower left quadrant, showing relatively lower levels of diversity and cohesion, and mostly lower activity, were largely closed groups; and should perhaps re-visit their worth on Workplace.
Taking a Deeper Dive
In our sample, interestingly, the Open groups outnumbered the Closed groups by 75 to 66. While on the surface, the above analysis might suggest that Closed groups are better than Open groups, all is not as it seems. Firstly, the group size will also have something to do with performance. On average the Open groups had some 375 members, while the Close groups averaged only 67 members. Most of the high performing groups in the chart above were relatively small groups, compared with the Open groups.
Groups size and Open/Closed status are just two factors that might impact group performance. At SWOOP we collect over 20 collaboration factors, some of which will influence group networking performance more than others. After correlating each of these measures against our performance measure, we found six factors which showed a strong correlation with performance. Two of the factors were negatively correlated:
*Red bars indicate a negative correlation
The following table describes each factor:
|% Observer||The proportion of group members that have been active less than once every two weeks.|
|Key Player Index||The Key Player Index is a measure of how large and interconnected the leadership of the group is. In groups with many leaders, the Key Player Index will be high. Where a group relies heavily on a single leader, it will be low.|
|Activities/User||The total of all posts, likes, replies divided by the number of active users. The Activity/User measure is a simple indicator of the spread of activity across the group. The higher the score, the more likely that members are equally engaged in the group’s activity.|
|Threads/User||The total number of discussion threads divided by the number of active users. The Threads/User measure is used to gauge the depth of discussions being had. Longer threads infer deeper conversations. When spread across the membership, stronger group engagement is inferred.|
|%Engager||The proportion of non-observer members who effectively balance contributions made and received.|
|%Catalyst||The proportion of non-observer members who are the target of more contributions compared with the contributions they make e.g. a typical senior executive.|
The %Observers and %Catalyst were found to be negatively correlated with performance. The other factors not referenced here may still be highly relevant, but didn’t correlate to high levels of diversity and cohesion at the group level. We will now discuss some of these measures in more depth:
The Behavioural Personas
The negatively correlated %Observers is no surprise. If you have a group high in Observers, it can be hard to generate a strong collaborative performance. The %Catalysts is somewhat of a surprise, as in general, we see Catalysts as a positive persona for driving activity. Perhaps once a certain level of activity is achieved, then the Catalyst may become more of a liability, as groups look for the more balanced interactions, as demonstrated by the Engager persona. The aspirational Persona is the Engager, who typically is seen as a leader who can broker connections and therefore engagement amongst the group. The results support our hypothesis that the Engager is the persona that is most associated with higher group performance.
Key Player Index
Groups with a high Key Player Index have a high number of people connecting the group. Groups with a low Key Player Index rely heavily on a selected few people. Being very reliant on a low number of connectors obviously means more risk and is less desirable.
The graphic below illustrates the extremes of the KPI measure:
Therefore, if you want a high performing group, you must encourage the group to develop many interconnected ‘leaders’ to sustain growth and member engagement and guard against single points of failure.
Activity is the count of all posts, replies, likes and mentions. Members are the count of active participants in the group. Activity levels were positively correlated with performance (0.29), while the number of active members was negatively correlated with performance (-0.24). As a ratio therefore, smaller more active groups are seen to be most highly correlated with performance. It is important to note however that the %Public (Open groups = 100%; Closed Groups = 0%) had close to zero correlation with performance. So the closed groups that are high on the performance chart are more likely there because of their relatively smaller size and high activity, than their Open/Close status.
Summarising our Insights on Workplace Groups
There is a lot to take in on this initial analysis of Workplace groups and the impact of closing groups. Here are the key points:
- Whether a group was closed or open showed no correlation with performance. It’s more likely the smaller size of the group and the level of activity that contributes more to performance. Use the Activity/User measure to monitor this.
- Observers may be gaining some personal benefit from joining groups; but on balance, the group will benefit more from their active participation.
- The behavioural persona most associated with performance is the Engager. Groups that have a high proportion of Engagers amongst their leadership are more likely to perform better. Individuals should aspire to the Engager persona.
- For those classified as Catalysts, monitor the maturity of your group’s activity levels. Once activity has reached a sustainable level, look to move toward becoming an Engager, by consciously responding to group members that respond to you.
- One of the most influential factors correlated with performance is the size and interconnectedness of the core members of the group. Groups should aim to build a strong inner core of leaders, to the extent that there is no reliance on single individuals for group success.
- The density of group discussions can be indicated through the length of the discussion threads and the breadth of group members involved in them. Deeper and broader discussions infer stronger group engagement in key group issues. Use the Threads/User measure to monitor this.
- Overall success will be dictated by group purpose. In the absence of a context where a group only aims to be diverse, or only be cohesive, it is better to maximise both.
Should you opt for an Open or Closed Group in Workplace?
We think that at the early stages of adoption, the Workplace recommendation for closing a group to prevent ‘damage’ from leakage of private messages to the news feed is sound. Attracting adopters can be hard enough, without having to deal with this added risk. We would anticipate however, that over time, the focus on the general news feed will diminish as people learn to go to their groups first, for their more important interactions, rather than the news feed. This was certainly the case with other ESN platforms.
Closing a group, however, does not have to mean no knowledge sharing. As our results have shown, knowledge is best shared person to person. If a closed group has members that only interact in that group, then it is a true silo and should be avoided. However, if the group has members who are active in many other groups (high personal diversity), then there is a high likelihood that the members will be able to share knowledge gained from their participation in these other groups; and vice versa. The ‘holes’ in the silos are in fact the cross group participation of the members. Arguably, this style of knowledge sharing is even better than simply opening up all information resources; as knowledge can be shared in context and in person. It is therefore important to welcome and indeed encourage multi-group participation by your members; in particular, if your group is closed.
In the longer term, everything else being equal, we still believe that Open groups are preferable to Closed ones. Culturally, it sends a better message, supporting the principles of open knowledge sharing and working out loud.
We have provided a single group performance measure here that calls for maximising both diversity and cohesion. We do understand, however, that this is not always the case. Just because an open group is not cohesive, should that matter? What about closed teams; does it matter that they are less diverse? Your context might be a retail environment, and you only want your shops to care about cohesion among the team members in the store. Is that a problem?
We are only now scratching the surface of the insights we can draw from analysing collaboration in Workplace. The results presented here were drawn from a single organisation, although it was a large and active one. They may not hold true for other organisations, so we’ll continue analysing this. Our analyses rely on simple correlations; and we know that correlation does not necessarily infer cause and effect. Stay tuned for our series of follow up posts, where we will discuss these issues and more.
Want to find out the group cohesion and diversity for your Workplace groups? Try SWOOP Analytics today.