Organizational Diagnostics

Why Employee Surveys Often Fail

Return to Articles listing

Most of the 200+ companies we have worked with over the last 20 years have shared a horror story of an employee survey effort that went sour.  Despite the best of intentions, it left employees bitter and disillusioned, and left executives shaking their heads and wondering how it all went so wrong.

The steps to a well-intentioned but ultimately disastrous employee survey are easy to list.  Unfortunately they are so innocent and rational that many companies have already dug themselves in too deep to recover before they even realize they are even in trouble at all. In the hopes of helping you avoid your own horror story, here's what NOT to do in building a survey:

Identify the issues on which you would like more information.

Typically HR staff and executives brainstorm on the questions they would like to pose:

• Are employees happy with the performance of their immediate manager?
• Are employees content with the compensation and benefits available?
• Is the recent acquisition going well?
• Are employees optimistic about our strategic direction?
• How do they rate their senior leadership?
• Are the values we rolled out last year being used?
• Do they think the downsizing was handled as well as could be expected?
• And so on and so on.

Expand the list from STEP 1 into a survey instrument.

Any one (who believes they are) fluent in survey design could expand on the above listing to create an actual instrument.  Most likely they throw in a few open-ended questions to catch any issues not included.

By now it looks like they have a perfectly good survey!  It covers the questions everyone had and provides employees a chance to speak to the critical issues of the company.  What else could we want?

Enter and analyze the data.

For some companies they took the foresight to purchase a survey analysis software package so they are able to smoothly input the data and pull off the most obvious reports.  For others, there is a last-minute scramble to handle hundreds of surveys and high volume of comments.

Present the data back to the organization

Now the whole process starts to fall apart!  Reports are requested from every quarter of the organization.  Managers want to see "their data" without the distraction of other departments or divisions.  Some people want horizontal bar charts, others want mean scores, others want polar plots.  Some people want to see standard deviations and skew, while others want "just the highlights". People start to ask for unusual cuts:  "Can we see just the union employees at the Oregon site who also indicated they were unhappy with our benefits program?".

People start speculating on using the data in ways no one ever anticipated.  Couldn't the average scores by site be factored into the performance evaluation of the site managers?  Shouldn't managers be accountable somehow for the morale scores for their groups?  Suddenly there are key players becoming defensive rather than curious.

The poor staff in charge of analyzing and reporting out the data find that they are inundated with essentially clerical tasks, with little bandwidth for working through the data in any intelligent, sophisticated way.  Data roll out is complicated by new politically sensitive options.  A few statistically sophisticated employees begin to complain that the data analysis is not being handled well, or that key technical issues in sampling and question construction were ignored.  People begin to pick at the design of the survey (especially site managers and department heads), challenging the phrasing of questions and the intelligibility of the data.  The debate starts to center around the survey process rather than around the implications of the data.

We will explore in other articles how to avoid this common and disappointing outcome.  For our purposes here, it is enough to summarize the underlying flaw:

Surveys are too often viewed as data collection efforts (after all, we'll be collecting a lot of data, right?).  In reality, employee surveys are decision efforts, with an inordinately large data collection effort at the beginning.

But it is the decisions made with survey data that justify the entire effort.  The appropriate focus for the design effort is NOT "what data to collect", but rather "what decisions are we hoping to allow by collecting data? "  The initial focus needs to be on building the decision readiness of the organization. When the data collection is divorced from its ultimate use, there are few constraints and direction.  The inevitable result is a loose assemblage of data that is more accusatory than illuminating.  Without a clear sense of how the data is meant to be used, managers and staff imagine the many dangerous ways it might be used.