HR is becoming more and more data based and analytical. Yet the insights that we’re getting from the data are not increasing as quickly as the rapid proliferation of software tools, seminars, and people with “data scientist” titles. Why is that? A big part of the problem lies in the kinds of analysis we’re doing, with too much emphasis focused on techniques that appear to provide new insights, but which more often than not are a distraction: data mining and linkage analysis.

Data mining. The motivation behind data mining is eminently reasonable: you look for statistically meaningful relationships between measures to inform further analysis. However, data mining is one of the most risky things you can do with HR data. If people are clamoring for insights and you find an interesting relationship in the data, it is tempting to present the results to engage your audience. But presenting interesting relationships is very different from specifying and testing a full causal model.

The risk is that if you show them a statistically significant relationship, they likely will assume something should be done. Action-oriented leaders, who are commonly found, often jump to designing solutions based on bits of data and preliminary but incomplete insights. They figure that doing something is better than doing nothing.

For example, suppose there is a positive correlation between performance ratings and people who are promoted from within, compared to people hired in from outside. A logical conclusion is that internal promotions are more productive. After one look at that simple analysis, an action-oriented leader might easily conclude that external hiring should be deemphasized. But further analysis might undermine or even contradict that conclusion.

Performance ratings can be biased by cultural fit. Internally promoted people may get higher ratings because they know how to act within the company culture, even if they aren’t more productive. External hires also can have a harder time knowing where to go to get help. So they may indeed have lower performance, but not because they are incapable of performing—it’s because they aren’t properly supported by and integrated into the organization.

Both of these possible explanations are reasons why it is premature to jump to a solution like emphasizing internal promotions. First you need to establish the larger business need addressed by hiring externally. Fresh blood from outside may be critical for staying current on customer trends and your competitors’ strategies. If those fresh faces get lower performance ratings because they challenge the dominant culture, that may be because they are doing exactly what you want them to do.

Linkage analysis. Linkage analysis has received a lot of attention in recent years. It offers the promise of demonstrating a connection between investments and process improvements on the one hand and operational metrics on the other hand. It consists of statistically estimating the relationship between changes in the investments or process improvements and changes in the operational metrics.

In its simplest form, linkage analysis connects measurements involving people (employee attitudes and capabilities) with business performance metrics. Within the scientific community there is considerable debate whether the positive statistical relationship measures true causation or just correlation. For example, those firms that are the most successful also have greater means to spend more lavishly on their employees—Google is the most prominent current example. A statistical link between HR interventions and business performance can exist even when the causation runs from business performance to HR spending, not vice versa.

Linkage analysis is appealing because you can always construct an argument for why the measurements involving people should matter for business performance. In many cases it is likely that the causation does run from spending on people and HR to positive business impacts. However, just because something should contribute to improved organizational effectiveness and performance, there typically is no guarantee that it will—and often many scenarios under which it won’t. For example:

  • Training plays an important role in closing competency gaps, so measures of training incidence should be correlated with improved performance; however, training is not the only contributor to improved performance and often not the most important one.
  • Measures of competency demonstration should also be correlated with improved performance, even though some of the most important competencies that drive performance cannot be measured well and thus are missing from the set of measurements.
  • Coaching is an important part of effective feedback and performance management, yet it is rarely cited as the most important or critical barrier to performance; instead, it is a contributor to improved performance while rarely, if ever, being “the” cause.

The problem is with linkage analysis that focuses on only one of the measures above. The analysis is highly likely to show a statistical correlation between the people measure and organizational effectiveness and performance. As the discussion above demonstrates, even though the correlation may be strong, it cannot be attributed to true causation.

A different approach. The common problem with data mining and linkage analysis is a missing link to the questions that matter most for the business, and the real causes of engagement and performance. There are literally hundreds – often thousands – of things you can do to try and improve organizational effectiveness. You can spend your entire career wading through massive amounts of data and analysis, finding lots of interesting statistically significant relationships that may help improve bottom line results. But to find the greatest return from making changes to people and processes, you have to take a more systematic look at what are the barriers to and enablers of improved strategy execution.

To get there, you have to start with building and testing causal models of organizational performance. That requires integrating the business and human capital approaches to analysis, versus what happens all the time today: enterprise analytics and HR analytics are conducted separately, leading to more dead ends, false positives, and journeys down rabbit holes than anyone could ever imagine.

There are three steps in an integrated approach:

  1. Start with competitive advantage analytics, analyzing the organizational capabilities that enable your business to make money. Where is your organization strong versus weak in its competitive positioning in the market, and where can the biggest gains be made?
  2. The second step is enterprise analytics, looking at the elements of organization design and culture where there are the greatest opportunities for improvement.
  3. The third step is human capital analytics, focusing on the individual (role) capabilities, job design and motivation factors that can get in the way of effective performance.

When done right, following these steps almost always leads to deeper and more meaningful insights that the business can leverage to really improve strategy execution and organizational effectiveness.

Leave a Reply

Your email address will not be published.