There is a growing acceptance with Employee Engagement that how we measure and ultimately manage it needs to change. Studies like that published this weekby ISS and Morten Kamp Andersen show strong links between Engagement and business outcomes.
In his article in Forbes in August Josh Bersin stated that Engagement was a $1bn industry. Business has probably been measuring and managing engagement for about the last 10 to 15 years. Given the spend and the duration we’ve been focussing on it you’d have expected that Engagement levels would have improved. Instead they’ve remained stubbornly flat at best.
To give some explanation why this may be the case I want to use a simple dynamic model to show how engagement levels develop and what managers can reasonably do to effectively manage engagement within their firms.
What is systems dynamics?
Systems dynamics is a modelling approach which enables you to understand the non-linear behaviour of systems over time using stocks and flows. Much of what we want to understand in workforces can effectively be understood with dynamic systems. For example, in its simplest state an organization could be thought of as a stock of employees with a flow in (hires) and a flow out (leavers).
The historic way of understanding engagement has been to take a snapshot, most often through the use of a census / survey at a certain interval, typically annually. The results are then aggregated and then compared with previous snapshots or populations.
As well as being far too infrequently measured, by using aggregated snapshots we see proportions at each time but don’t get to understand how changes occur. We have no way of knowing whether the results come from the same or different people and whether these individuals are being engaged or disengaged. There is a big difference between a function which shows 50% engagement during two consecutive periods because the same people are engaged to one where everyone swaps between engagement and disengagement or because different people have responded. The current approach does nothing to show this.
Engagement as a dynamic system
Instead of viewing engagement levels as a static model it is helpful to think about the process of how engagement levels develop over time.
The simplified model below was developed using a modelling tool called Sysdea. I’ve created two populations in this imaginary firm; engaged and disengaged employees.
New employees join the firm through hiring. From experience, data shows that new hires are almost all engaged when they start (we see 98%+ levels of engagement at 3 months in clients). The model therefore has hiring as a flow of engaged employees into the firm. For simplicity I’ve added a rule that #hires = #leavers in any period (i.e. the firm always has the same number of employees)
If we think that all new employees are initially engaged, but that the firm has a lower level of engagement (eg 64%) then there must be a flow of employees from the engaged pool to the disengaged pool. This is determined by the disengagement rate – what proportion of engaged employees become disengaged each time period. Conversely we hope that we can re-engage employees who were previously disengaged, described using the reengagement rate.
One of the cost implications of disengaged employees is that they leave the firm at a greater rate than engaged employees. I include this using different leaving flows & rates for engaged and disengaged employees.
In the above simulation run of this model we start with initially all employees as engaged (let’s say we hired them all at the same time). I’ve added 100 employees which means that the numbers of engaged employees at any period is the engagement rate.
As you can see from the line graph inside the engaged employee stock the group quickly converges to around 64% – it’s at 67% within 3 years and at 64% at 5 years. (stocks actually show the number but as the total is always 100 the engaged stock can be read as a percentage).
What happens if we start with different stocks of engaged and disengaged employees but keep the rates the same?
In the second simulated example I’ve used the other extreme case – where we start with all disengaged employees. Again we see convergence to exactly the same level at around 5 years.
This result might be non-intuitive – systems dynamics models frequently are – but demonstrates an important message. The only way of changing engagement levels in the firm is by changing the rates. If we were to change the stocks at any time – for example by a short-term or one-off initiative – you’d quickly get convergence to the same level unless you change the long-term rates.
Originally I had stated that 20% of engaged employees became engaged each period and 5% of disengaged employees became reengaged again. If we halve the disengagement rate from 20–10% then firm-wide engagement levels converge to 77%. If instead we double the re-engagement rate from 5–10% we only see a growth of engagement to 67%.
The other way we could conceive of increasing firm-level engagement is by encouraging more disengaged employees to leave. In our model doubling the disengaged leaving rate from 30% to 60% of disengaged employees leaving gets us to around 77% firm-wide engagement as well, however doing this produces 16% overall staff turnover compared to 8% with the disengagement rate at 10%.
If this model shows one thing it’s that the most effective way to change the overall level of engagement in a firm is by changing the rate at which you’re disengaging employees.
We know two important lessons from studies of customer engagement – that the experiences an individual receives are the big difference between engagement levels and that certain events have more impact on engagement than others.
Assuming this applies to employee engagement then the key thing we need to understand and act on is not how to engage our workforce but how to stop disengaging them by focussing on events and experiences that contribute to the disengagement rate.
The traditional approach to measuring engagement focuses on the difference in perception between engaged and disengaged employees at a single point in time. By considering engagement as a dynamic system we can see what’s important is to measure events and experiences that cause an engaged employee to change to a disengaged employee.
Workometry as a way to understand engagement / disengagement rates.
When we were designing Workometry, our employee feedback solution, we started with this dynamic model as a foundation. Therefore it was important to provide a data structure that made longitudinal analysis – how an employee changes from engagement to disengagement (and back) over time – easy. Workometry can be thought of as a tool to help capture the engagement / disengagement rates and the experiences that trigger changes.
The other aspect that we included was the ability to use existing data about events – for example those stored in HR systems – within the analysis. We should use available data where possible. Most events are captured on some system. It makes sense to combine this with the perception data for analysis.
The huge benefit from understanding events and experiences that determine the engagement / disengagement rates is that it’s much easier to develop effective actions. We can be much more specific & focussed in what we do because the diagnostic data is less ambiguous. Top-down action plans are almost entirely obsolete and changes can be much more rapid.
What is not included in this simplistic model is how different people are affected by an event in different ways. However, the systems model, and of course our analysis, can incorporate this information. Adding it does it fundamentally change the high-level implications but is important when prioritising actions. We can integrate external data such as local employment rates to help understand how the economic lifecycle will change firm-wide engagement.