Trusted Partners or the Machine?
In our first class this semester, back in the halcyon days of our pre-quarantine innocence, Professor Polzer posed a question: Are employees trusted partners in a mission or cogs in a machine? Though we have discussed and debated, pondered and premised for almost a full semester now, these authors feel no less conflicted as to this central tension of people analytics than we did almost three months ago.
“It should be about trust, babe (Be about trust)” – Ashanti & Ja-Rule
Introduction
In our first class this semester, back in the halcyon days of our pre-quarantine innocence, Professor Polzer posed a question: Are employees trusted partners in a mission or cogs in a machine? Though we have discussed and debated, pondered and premised for almost a full semester now, these authors feel no less conflicted as to this central tension of people analytics than we did almost three months ago. Indeed, further reading has prompted even more questions: Are People Analytics fundamentally different to other big data disciplines (People Are Not Cogs)? How actionable are our insights (People Analytics Must be User Friendly)? And what is the role of trust in this space (The Enemies of Trust)? These authors would suggest several points to further these discussions:
Models are a simplification of reality, but that doesn’t mean they don’t have value.
Initially, there seems to be a contradiction at the heart of our class discussions. We talk about complex human systems and employee-centricity, yet build simple (relatively) mechanistic models that treat employees as mere dependent variables, nodes, or data points. Yet, models are always abstractions of reality. It’s impossible to capture everything–we either lack the data, the understanding, or the computational power. In weather forecasting, economic modelling, or even epidemiological studies, we always simplify reality.
The value of models isn’t that they are perfect representations of reality, it’s that they help us understand underlying relationships and focus our attention on what is most important. But this abstraction requires us to adopt a healthy dose of humility in our work. Any system involving humans will be subject to human complexity. Feedback loops, unintended consequences, and difficult-to-quantify factors abound. In People Analytics, perhaps more so than in many other data-based fields, we must be cautious in how we draw conclusions or proscribe action.
The employee-employer relationship is two-sided–both parties must want to be trusted partners before such a relationship can emerge.
Trust is a critical component of the relationships an organisation has with its employees. We want our employees to be trusted partners in a joint mission–certainly we do in modern knowledge work where we need employees’ creativity and engagement to co-create value–but employees have to want the same before it can emerge.
Currently, most organizations utilize people analytics to measure and address employee engagement, productivity, and other “efficiency” metrics. Though valuable, the evidence linking employee engagement and productivity to overall organizational performance is mixed, or at best, incomplete. But what is clear, is that employee trust is a strong leading indicator of organizational performance in a variety of sectors around the world. Measuring levels of employee trust can provide insights that help employers understand future retention patterns, productivity risks, and early warnings of issues festering beneath the radar.
But trust is a complex creature. It’s tougher to measure than simple employee engagement–employees may be outwardly engaged and productive, but internally disengaged due to a trust deficit. It’s also a function of a wide variety of employer-employee interactions, built up through communications, impressions, transparency, recognition, and thousands of other small details.
Consequently, the need for organizations to develop better methods of measuring and understanding levels of internal trust has never been greater.
Trust is traditionally measured through organizational surveys, but most employee engagement surveys do not ask the right questions to accurately determine levels of trust. Good questions on organizational trust seek to measure the competence, integrity, and reliability of an organization’s stakeholders (e.g., direct managers or senior executives). But such surveys must be carefully constructed to elicit the right information from employees. Given that poor communication and fear of retribution are often the consequence of trust deficits within an organization, an appropriately designed organizational trust survey should provide respondents with anonymity. Data collected from survey questions like this can be further analyzed by employee tenure, location, compensation, age, gender, etc in order to increase the depth and variety of insights that can be gained.
As people analytics continues to be an ever-present resource for organizations in accomplishing their performance goals, we hope that researchers and practitioners in the field increasingly prioritize the creation of actionable and quantifiable systems for measuring and improving organizational trust.
I agree Trust is important and that it requires reciprocity on both sides.
Employees are often skeptical of filling multiple surveys without know what management is thinking. As society grows more weary about organizations gathering and tracking data, increase in surveys, though anonymous, could also weary employee and put more strain on the little trust that exist.
One way I believe organizations can push the boundaries of trust further is by leveraging the set of traditional organizational surveys to collect anonymous data from (not only employees but also) from the management staff. These data could then be analyzed and insights could be shared with the employees.
This approach can also help open communication lines and help both parties communicate their feeling and priorities. This could also help remove the stigma of being monitored by management when they realize the everyone is being surveyed.
Toni, I loved your hook about “the halcyon days of our pre-quarantine innocence.” How I miss those days!
Your point about the importance of trust is a very important one. You mention the need to measure levels of trust, which I agree can be very helpful. I also wonder whether organizations can apply theory about the *structural* determinants of trust in an organization (I imagine such a literature exists) as a complement to their people-analytics efforts.
For example, I have read a bit about how in Germany, strong labor-management relations have been a key contributor to the country’s economic success. Structured negotiations between unions and employer associations help facilitate harmonious relations.
When it comes to people analytics, perhaps dialogue between groups representing different interests within the organization is part of the solution to the trust issue.
Interesting thoughts Toni. There’s a lot to unpack in your initial question. I’d say, for better or for worse, we’re heading towards an age where employees are cogs in a machine and the notion of being trusted partners will be a ‘luxury’ of sorts for only those in the upper echelons. The big tech companies of now might offer a glimpse into what to expect going forward.
While I agree that models are a redacted version of reality, I am not entirely sure that depending on human intuition will necessary lead to outcomes where there is more trust. Indeed there are situations where models (probably less in the people analytics space but I can think of other disciplines) can be definitely more trustworthy that human judgement. Maybe as people analytics becomes more sophisticated we might see a switch in this space as well.
Also on board that surveys, self reported or otherwise, are not a great tool to capture an accurate picture of what people actually feel or think. What’s somewhat frightening however is that with the introduction of more data collection tools (such as complex wearables or sensors) surveys might not be necessary in the first place.