Jeffrey Saltzman's Blog

Enhancing Organizational Performance

Employee Survey Interpretation – 103

with one comment

One way to help prioritize which findings from an employee survey the organization should focus upon has to do with the items from the survey that are found to “link” most strongly to various business or organizational performance metrics. These linkages can give guidance on where resources should be spent to have the largest potential payback to the organization.

We perform linkages in our mind on a routine basis to help us prioritize all of our activities, to help us direct our energy, and hypothesized linkages are rigorously debated publicly as policy matters. For instance one debated linkage has to do with the playing of violent video games giving rise to increased violence in the behavior of children. What about the effect of watching violent cartoons or movies? Will the increase in troop strength in Iraq, the “surge” help to stabilize the situation? One group hypothesizing a linkage feels it will, another group says it will have no or little effect. Even after it has occurred the hypothesized linkage is not always clear as many other factors, other than the surge itself, come into play. Will the tax rebate that is being prepared for distribution this May actually spur the economy? Is there a link? Or if the money had been used in another fashion would the positive effect on the economy be greater? Because of the large number of moving pieces, the number of variables to be taken into account when answering that question the ultimate answer is unlikely to be exceedingly clear.

This same kind of linkage logic in a simplified fashion can be used in helping to determine which employee survey items action should be taken upon. Will improving on speed of decision making have a greater impact on organizational performance than providing additional training for staff on how to perform their jobs? What if decision making was moderately favorably rated but training was poorly rated? Which action would have a greater impact in that situation? What if the ratings were reversed? It can get rather complicated, but it doesn’t need to be.

Linkages can be done at a single point in time (a concurrent study) or they can be done over a period of time (a predictive study). A third approach is a bit forensic in nature and that is to look at outcomes (e.g. find all organizations that had an increase in stock price of at least 10% in 2007 vs. those that had 0% or negative returns) and then trace backwards to what environmental/organizational factors impacted them (a postdictive study). By far the easiest way and fastest approach to applying linkage findings is by studying the organization at the current moment in time as there are many fewer variables to control.

The success of a concurrent study, examining an organization’s results at a single point in time, is dependent on their being variance by organizational sub-unit on the responses to the survey items and other business performance metrics. If a survey is designed as a “feel-good” measure or the response to an item is across the board very favorably rated (e.g. safety or ethics), yielding uniformly positive results among the various organizational units, the ability to successfully perform linkage in less. (See The Fallacy of Ample Parking).

The first step in a concurrent study is to obtain a one-to-one correspondence for each organizational unit for their survey results and their corresponding business performance metrics. So if we have survey results for 100 units (departments, divisions, plants, stores etc.), I want to obtain the corresponding business performance metrics of relevance for those units. Those business metrics can be related to personnel issues, quality, production, sales, financial performance, etc. and could include measures such as voluntary turnover, sales per square foot, accident or injury rates, absenteeism, plant availability or uptime, production goal attainment, error rates, shrinkage, etc. The number and type of metrics that can be used are limited only by what is collected and relevant to the organization.

The correspondence really between the survey results and the business metrics should be one-to-one with no over lap – so that each observation or line of data (both survey and business metric) is mutually exclusive of the others or orthogonal in nature. Say you had a retail chain with 100 stores. If either the customers shop at or the employees work at more than one of the stores, the one-to-one correspondence suffers and it may be better to perform the linkage not at the store level in this case but at the region level, as it is much more likely that both employees and customers who may interact at numerous stores are more likely to stay within the region.  The ideal case would be to have all employees only work in one store, to have all customers only shop in one store and to have other business metrics (turnover, shrink, etc.) cleanly measured store by store. It makes sense if you think of what the goal is – to determine which employee attitudes correspond to which customer perceptions or other business metrics.

It is often difficult to conduct linkage work across divisions of an organization in one study as the performance metrics, even when they are given similar or identical names, may be measuring different things or have different acceptable levels. A score on a particular metric of 75 in one division may be a very good score and that same score on the same metric in a different division may be only average. In other words, the same score can mean two different things across a broad piece of an organization and that will prevent a successful linkage from being carried out. Conducting linkage with smaller pieces of the organization increases the likelihood that a 75 is a 75 is a 75 and that they all mean the same thing.

Many times when conducting linkage work the greatest effort or amount of time is spent getting the data in order or understanding what is actually in the data for the study to be successfully completed. Once the dataset is cleaned up and understood then the analyses can commence. The analyses used to study these “linked” datasets can be quite varied and can range from simple correlations to structured equation modeling to a whole host of other techniques to tease out meaning. Invariably the chicken and the egg question get raised. Do you need conditions that generate an excited, engaged workforce to obtain great business results or do great business results allow you to create conditions that get employees excited and engaged? The answer is you need both as one does exclusively cause the other and they feed off of each other in a reinforcing fashion.

In the hundreds of linkage studies that I have reviewed or participated in, almost every time the case can be made that certain employee attitudes can predict to customer attitudes and other business performance metrics. In the few cases where it did not work, it is invariable due to the issues within the original dataset (e.g. what are called the same variables across an organization are actually being used to measure different things, the lack of mutually exclusive data points).

While linkage can be time consuming and difficult to conduct it often times yields extremely compelling information that can help guide an organization’s decision making regarding which items from a survey to focus resources and attention upon.

Written by Jeffrey M. Saltzman

November 16, 2009 at 8:06 am

One Response

Subscribe to comments with RSS.

  1. I agree that the hardest part of this linkage is often getting good business metric data to match up to the employee survey data

    Ed Nichols

    April 29, 2010 at 11:18 am

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: