Highlights

The analysis involved observing the level of awareness and interaction between teams on key pieces of communication from the top, or senior management. This identified previously unknown information flow bottlenecks and communication deterioration amongst certain teams.

This was based on understanding the invisible social networks of communication within organisations rather than just the formal hierarchical structures. Up to 70% of knowledge flow can take place through these organisational social networks.

Highlights

We initially deployed GalaxyScope to analyse the communication patterns of the general global public (not employees of the company) in relation to the specific morals, ethics and accountability issues relevant to our client. This involved the identification of virtual tribes on the web which are groups of people who share the same values and collective awareness. Sources include social media platforms like Twitter, Wikipedia and subject-specific forums.

This provides a very rich source of machine learning that enables algorithms to develop a much higher level of predictive accuracy when identifying the same behaviours within organisations, with naturally smaller data samples compared to the web. The communication characteristics of the identified virtual tribes were then compared to the communication attributes of our client’s workforce to identify relevant tribes within its organisation.

For example, this included the identification of internal organisational tribes on the basis of the Schwartz Theory of Moral Values (a well-known model for ten universal human values). The virtual tribe analysis conducted through GalaxyScope was supplemented with an analysis of internal client communication patterns and social networks using GalaxyLens.

Highlights

The inclusion of content related data, such as analysing the type of sentiment in communication, can understandably raise concerns regarding individual privacy. This can be effectively addressed by encrypting the data at the source (thereby rendering the content indecipherable) and aggregating the reporting at team level (typically, 10 people and above) so that individuals cannot be identified.

When these security and privacy controls are applied, the results from the analysis are reported on the basis of summarised and anonymised team patterns, which protect privacy whilst providing valuable insight into the sentiment and level of psychological safety being experienced by teams. In this case, our analysis focused on the ratio of positive and negative sentiment.

Perhaps somewhat counterintuitively, a disproportionate amount of positive sentiment can reflect an increased level of fear. When people fear speaking up or calling out bad news, the general sentiment is typically skewed towards the positive. Whilst there is nothing wrong with positive sentiment per se, an appropriate balance of positive and negative sentiment means that people are more likely to speak up because they are also in the habit of expressing concerns. This balance of positive and negative sentiment reflects the use of “honest language”.

Highlights

Our technology is not intended to function as a strict surveillance tool. It's real value is in providing highly accurate measurement and prediction of areas of the business that may be more susceptible or predisposed towards misconduct. This enables the organisation to proactively intervene in these areas before material problems arise.

In this way, our technology is designed to prevent misconduct before potential bad actors have the opportunity to engage in unwanted behaviour. 

In this case, we used a unique combination of social network analysis to identify hidden pathways of  potential collusion, dynamic analysis of communication patterns to reveal behaviour consistent with misconduct and content analysis to demonstrate language reflective of unwanted behaviour.

This included a back-test of known bad actors within the organisation. The intention of the back-testing was to help calibrate our algorithm for this particular client's environment and demonstrate the predictive accuracy of our analysis.