My colleague David Pitre gives this advice: If you are interested in learning more about employee communication measurement, use every opportunity  to take a survey or participate in focus groups. Then observe what works or doesn’t work, and apply what you’ve learned to your own measurement efforts.

It’s great advice. The other day I found myself with a few minutes to spare, so I decided to take an electronic survey for Fast Company magazine. The survey turned out to be a train wreck, but I learned something.

Early on, the survey had some annoying features—for example, there was no way to tell much progress I was making—but I trundled along until I got to this question:

“In which of the following categories are you involved in the decision-making process at your company?”

Underneath, there was a list of about 20 categories, such as financial services,
real estate site selection and insurance services, and instructions to “select all that apply.” For each category, survey takers could choose between three responses:

• Determine the need/define requirements
• Evaluate/recommend
• Authorize/approve

So, for example, at my company I “authorize/approve” insurance selections, so I made that choice.

The trouble was, some of the choices didn’t apply. I don’t get involved at all in some decisions, or my company doesn’t buy those products or services. So I left those answers blank, as I did the last two categories: “Other” and “None of the above.”

That’s when the trouble started. When I clicked on “Continue,” I got an error message: “Please select as least one option per column.” Annoying, because the available responses didn’t apply to every category, so I had to make selections that weren’t accurate. I tried to skip it. But when I got the error message again, I thought, “What the hell!” and selected a choice for every category except “Other” and “None of the above.”

No luck. “Please select as least one option per column,” the error message read again. I carefully scanned the list. It seemed that the only way to advance would be to put an answer next to “Other” and “None.” That made no sense, but I answered “Authorize/approve” just to move on.

Now this error message came on: “You have selected None of the above, so please deselect other selections.” Aarrgh! I gritted my teeth, and decided to spite the survey designers (They’ll never know my true answers? Bwa-ha-ha!) by deselecting all of the other selections, leaving “None of the above” selected.

Nope, didn’t work. “Please select as least one option per column” read the error message. I was now steaming. For a few minutes, I tried different things to get the damn question to advance, but nothing worked.

That’s when I gave up. Aborted the mission. Closed the browser and went on my way.

If you are involved in employee communication measurement, here is why you should pay attention to this sad little episode. First, this was a missed opportunity. I wasted my time, and Fast Company lost the data it sought. But more important, because of my frustrating experience, I ended with a more negative opinion of Fast Company (and its stupid survey designers) than when I started. Remember that a survey is not simply a way to collect information, it’s a communication event that can shape people’s perceptions, either positively or negatively.

The lesson? Test your completed survey. Then test it again. Then ask someone else to test it. Only invite employees to take your survey after you’re sure it works.

That way, you’ll end up with a smart survey, not a stupid one that makes people cranky.