a network map of Microsoft

Graph AI for organizational analytics

New advances in graph machine learning paired with telemetry unlock a disruptive new ability to measure and reason about how organizations function. 

The modern workplace has changed the way collaborative work is performed.  With the shift to remote and hybrid work, more communications between people moved from conference rooms to digital platforms.  Using these aggregated signals from these digital platforms, we can create and deploy new technologies to drive better understanding as to how teams work together irrespective of the formal org chart.  Furthermore, by applying state-of-the-art advances in Graph AI, we can now empower every individual in these new workplace modalities to better support them in their collaborations by providing better recommendations for a wide variety of organizational tasks. 

Using these methods, we can see beyond the existing org chart – the limitations of org structure can be overcome by the reality of working relationships.  This also means we can move to models that dynamically adapt to the real-time interactions and relationships within the organization. This would allow companies to better understand the strengths of their teams, identify bottlenecks and inefficiencies, and make more informed decisions about resource allocation and talent retention. Additionally, graph AI can be used to analyze and predict patterns of collaboration and communication across the organization, providing valuable insights into how to optimize workflows and foster a more empowered and engaged workforce. Ultimately, the combination of the shift towards remote and hybrid work and the advancements in graph AI creates a unique opportunity for organizations to gain a deeper understanding of their operations and unlock new levels of creative output.

Using these techniques, we can imagine tools to solve a variety of tasks: from measuring the impact of mergers and acquisitions, to better quantifying the effects of reorgs, and to detecting the dynamic formation and dissolution of ad-hoc / project / or virtual teams (which often span org structure).  This last capability is especially critical as understating virtual teams spanning the org chart is a critical task for a modern organization.  Virtual teams often span formal organizational boundaries and may only exist for short periods of time as they are often formed for highly specific objectives on short time horizons (to which a formal org chart cannot adapt quickly enough).  Identification of virtual teams can lead to empowerment as those teams can be recognized and resourced.  This can provide a new set of tools that can be used by leadership as a signal to help drive alignment across the organization.  We are researching the new technologies to enable such capabilities, and much more, through the use of graph AI.  Already, some of our research has shipped within Microsoft Viva and we are continuing to explore new frontiers for new empowering technologies across the full suite of Microsoft products.

In addition to being able to run core analytics, we also want to ensure that users can meaningfully visualize and interact with the complex structures of an organization as a whole.  As such, we have also developed methods to automatically create Organizational Network Maps.  These tools, for the first time, provide a way to visually navigate an organization’s geography and terrain as it changes over time.  Additionally, these can be used to drive recommendation systems that make it easier to find colleagues and teams for new collaborations.  We can use these views to paint a vibrant picture of workplace behavior that doesn’t align with formal organizational structures.  Formal organizational structures alone have many shortcomings.  They can’t adapt at the speed of virtual teams, they are often determined with inconsistent methodology (each leader has their own optimal structure), and they are intensely hierarchical.  Creating new technologies that analyze collective behavior to automatically understand an org’s collaborative structure can help lead to improved systems that could ultimately help improve employee retention, facilitate space planning, and lead to better organizational outcomes. 

Finally, many of the graph AI techniques that we have been researching are directly applicable to other domains, such as counter human trafficking and anti-corruption.

More information can be found using the links below or by contacting Jonathan Larson.

Learn More: