It’s Time to Move Beyond “Digital Exhaust”

If you are a knowledge worker today, odds are you spend much of your day in a digital land characterized by a continuous flow of emails, Instant Messages, and video meetings. A land in which you might Slack the team members sitting across from you and 2000 miles away with equal frequency.

This “digital workplace” of today is unbinding us from physical spaces. It is changing the way we interact with our friends and colleagues. It is, increasingly, how work happens.

“Digital Exhaust” (What is it good for?)

Besides an ever-pinging smartphone, our continuous digital exchanges also produce what’s been characterized as “digital exhaust”: an immense and ever growing stash of the emails, chats and calendar events that are generated on a daily basis.

In workplaces heavily reliant on digital media for communication and collaboration, the immense value of this “digital exhaust” is gaining widespread attention as a tool to be utilized to understand where, when, and to whom information is flowing in an organization.

Over the past several years, organizational analytics researchers and industry practitioners have begun to tap into this rich source of data and extract value via a few extant strategies, ranging in maturity from academic projects to enterprise-grade products.

A few notable examples include:

• Microsoft Workplace Analytics (formerly VoloMetrix) ingests email and calendar metadata (fully ignoring message content) to provide work-life balance monitoring and tracking for digital signals which predict professional success. They mine a multitude of such signals, including individual focus time, frequency of messaging after work hours, hours worked daily, one-on-one meeting frequency, and the number of potentially bloated or inefficient meetings.
• Paul Leonardi, Professor of Technology Management at UC Santa Barbara, and Noshir Contractor, Professor of Behavioral Sciences at Northwestern University, jointly published an article in Harvard Business Review in late 2018 promoting the value of relational analytics in the people analytics space. Relational analytics is a strategy which goes beyond analyzing individuals as singular data points, to analyzing collaborative social networks. By analyzing communication patterns in organizations — who communicates with whom — they identified various structural network “signatures” which hold predictive value for individual roles (which employees are “influencers” or “innovators”) and team productivity (which teams are highly efficient).
• Sameer B. Srivastava, Associate Professor at UC Berkeley’s Haas School of Business, published a paper in 2017 wherein he analyzes actual electronic message content (instead of only metadata) captured over 5 years at a company of around 600 employees. The longitudinal nature of the data allows him to observe organizational enculturation or “culture fit” dynamically over time, proxied by the degree to which employees adapt and conform to the language use of their peers.

We applaud this ongoing evolution in analytical mindset and the technology which enables it.

We also believe that there is yet more value to be extracted from “digital exhaust,” and the key to doing so may be rethinking the concept of “digital exhaust” as a whole.

Limitations of “Digital Exhaust”

Up to this point, researchers and practitioners have primarily leveraged the metadata from “digital exhaust” in order to drive research. Metadata is the cursory (read: non-invasive) information that can be easily gathered from an email thread or calendar event in the form of  senders, recipients, and timestamps.

Most often, this data has been analyzed in a one-off, retrospective manner to be utilized by HR or organizational higher-ups. Generally speaking, employees also do not explicitly consent to their “digital exhaust” being used in such analysis (companies generally own their employees’ workplace emails and therefore are not required to ask for consent).

There is notably less research (although this is changing, see Srivastava above) that utilizes the more difficult to extract, though equally valuable aspects of “digital exhaust,” namely email and instant message content. There are both technical and cultural reasons to explain this lack of focus on message content, such as:

1. Natural Language Processing (NLP) techniques which empower researchers and practitioners to conduct rich analysis of text-based conversations have advanced significantly in the past several years.

 

 

2. Concern for employees privacy is likely a factor as well, or at least a concern for potential blowback that might occur should employees discover that email content is being analyzed without their consent.

Although “digital exhaust” metadata is highly useful for understanding social networks and influence, it only skims the surface of the social information available in the digital workplaces of today. With the advent of synchronous tools like Slack and Microsoft Teams, behaviors such as giving feedback and praise, disagreement and water-cooler banter, are making their way into our digital conversations. That is to say, our “digital exhaust” is increasingly an extension of our real-life selves.

So how might we capture the ever changing social richness of today’s digital spaces, and bring that value to employees?

We believe the answer lies in doing away with the term “digital exhaust” – a term which implies the static, emotionless output of work being done (metadata) – and instead viewing the digital workplace of today as a web of multifaceted digital relationships (content).

 

Opportunities for the “Digital Relationship”

As opposed to “digital exhaust”, a “digital relationship” is the full set of past, present, and future digital interactions between two employees in the organization. Scheduled one-on-one meetings. Late night Slack messages. Recognition emails. Metadata and the rich contextual data that make our digital workplaces so pivotal.

When we move beyond “digital exhaust” and begin exploring the “digital relationship” in this way, we find many new opportunities to better serve and empower employees, which will ultimately better the company as a whole.

How to use the “Digital Relationship” to Empower Employees

• Transparency is key in order to gain the most value from this data set. Employees should consent to their data being used in this manner, and those who consent should receive direct value.  Build the value for the employee first, and the value for the company will follow.

• By leveraging all accessible aspects of the digital relationship — both metadata and message content — we can gain a rich understanding of interpersonal, relational dynamics by applying existing natural language processing tools to extract content-based metadata such as politeness, sentiment, intent and language use frequencies. These dynamics evolve day by day, conversation by conversation, and as termed by Srivastava, constitute behavioral “trajectories” that can be tracked over longer periods of time.

• We can then drive value for employees by offering insights and interpretations of trajectories to managers and employees when immediately relevant, thus tightening the feedback loop. Managers may be made aware of many more content-based and metadata-based trends regarding the actions they take towards their team. This information can furthermore be contextualized by an understanding of the effects of their actions upon the immediate conversation, upon the productivity of their team as a whole, and upon the professional satisfaction and attainment of any direct reports. Thus, managers could simultaneously gain the opportunity to improve and change course before disaster strikes, but also the motivation to do so.

An Example of using AI to Empower Managers using their “Digital Relationships”

The “digital relationship” provides a unique opportunity to continuously quantify and track progress along a number of dimensions that affect the organization’s bottom line.

For example, organizational psychology research shows that it is important for teams to build trust with each other and create a “psychological safe zone” sans fear or judgment. To maintain such a safe zone, leaders in the organization should exhibit behaviors such as active listening and constructive disagreement. In a perfect world, managers and companies can directly measure the presence or absence of such behaviors. However, most existing digital quantification methods which might utilize “digital exhaust” metadata, such as determining how often an individual communicates with her team, do not wholly capture these abstract concepts.

Yet, with much collaboration in the modern workplace happening online, so we know that evidence of such social behavior (or lack thereof) is baked into digital conversations. For example, at Cultivate we utilize current NLP techniques to extract statistics from managers’ digital relationships with their direct reports, such as how quickly and thoroughly managers respond to requests for feedback, and whether managers tend to couch their disagreement in new suggestions. This fine-grained conversational analysis, turn by turn, gets us a step closer to measurement of the desired communication behavior itself.

By leveraging this conversation analysis in the digital relationship, an AI can then nudge managers to maintain awareness of their digital relationships and leadership skills, to revel in their proficiency in some skills and identify opportunities for improvement.

These personal insights give managers concrete reasons to seek improvement, whether it’s pointing them towards conclusions of optimal actions in the leadership research canon, or pointing out causal links between their leadership behavior and collaborative efficiency or leadership trust.

Furthermore, after a manager has made the decision to seek improvement, an AI is able to provide the information and the continued support they need to achieve behavior change. This includes sharing digestible leadership content in the form of daily tips and short videos, and pinpointing snippets of conversations that might have been executed more successfully.

With these advancements in AI, and those by other practitioners and academics in the field of people analytics, we hope to move beyond static, retroactive analysis of “digital exhaust.” Our digital workplaces of today are vibrant and ever changing, presenting their own unique challenges and advantages. We should be using equally dynamic technologies and frameworks to understand them, and we can start by recognizing that these are “digital relationships.”