Home Office: Design a Great Online Survey

Find out what your customers are thinking.

Steve Bass

Do you remember that SurveyMonkey survey I coerced you into taking back in April, the one where I asked your opinion about this newsletter and offered an all-expenses-paid trip to Hawaii? [He's lying about the free trip.--Editor] Well, the results are in; and it turns out all seven of you had lots to say.

Seriously, a little over 2 percent of you responded--about 2400 people. In this week's column I'll share the most valuable results, explain some of SurveyMonkey's best features, and offer tips and tricks for creating great Web-based surveys. Fair warning: This newsletter is going to be longer than usual.

Before Stepping Into the Survey

Hey, wait: You say you didn't read the original newsletter, the one about the survey? You might want do that so you know what I'm talking about.

You might also want to browse through the latest Home Office magazine column. In it I explain how SurveyMonkey works (and mention PollMonkey), talk about some fallout from the survey, and offer a few tips.

Dig This: Here's your chance to unroll a magic carpet and fly over any city in the United States. TerraFly is a dramatic example of how great technology can help you kill an hour instead of getting something valuable accomplished.

The Survey Sez... [Drum Roll Here]

My hunch is the first thing you'll want to do is check out the survey results page. (I know I did; so did everyone else at PCWorld.com.)

I'm jazzed about SurveyMonkey's power, and there are a couple of things I don't want you to miss about the product. (I know, I sound like an ad; nope, I have no investment in the company)

First, take a look at the variety of questions I was able to use: simple yes/no (question 6), "yes" with an essay option for "no" (question 8), and complex matrixes (question 9). BTW, don't get confused by the page numbers--they're in bold and are labeled "Untitled Page." Now find and click a "View" button (start with question 2). Here you'll find open-ended, or essay, results in which responders could mouth off, uh, say anything they want.

For me, this was the most valuable part of the survey--your individual comments. For example, many people said the e-mail subject line (the newsletter's title) didn't influence whether they read the newsletter. I found out most people wanted more "dig this" items, but quite a few readers didn't like or understand the phrase "dig this." Question 8, in which I asked if I should fiddle with the newsletter's content, was the most revealing. While lots of people were okay with what I say here, many others weren't too enthralled with my scribblings: "Less fluff, more content," "way too long," "less chatty, please," and "too many gimmicks" were just a few of the comments.

There's no doubt the survey's helped me refocus on what goes into the newsletter--and what stays out.

Before you move on, you might be interested to know I've created a survey for readers of my magazine Home Office column. You're welcome to take it (actually, I'd be delighted if you do). Of course, some of you might not have been reading my Home Office columns recently. No problemo--you can find a collection of them on PCWorld.com.

Watch a Video: We now offer video reviews of nifty products. That's right, you can get a glimpse of our senior editors smiling and sounding pleasant (oh, what they'll do for the camera). Watch Denny Arar get googly eyed at Hitachi's latest gee-whiz G1000 PDA Cell Phone.

Dig This: You've heard about Homeland Security's software that records every click of your mouse? Yep, it's a sophisticated tool that's completely transparent; most of you won't even notice it's running on your PC. It's still in beta, but I found a prerelease version you can try.

Get Some Survey Savvy

In my quest to design the perfect survey (no, I'm not even close), I did lots of research. I've come up with guidelines that will help you create your first survey. BTW, these tips aren't specific to SurveyMonkey.

1. Design, or roughly sketch out, your survey the old-fashioned way, with paper and pencil (don't panic, that's the long, yellow cylindrical object). It's much quicker and easier than creating the actual survey online.
2. As you think of things you want to know, always ask yourself: What will I learn, and what do I want to learn? Don't ask about anything that you're mildly curious about (like the questions I get from some of my editors) as too many questions will either bore or annoy responders. Stick with info that's vital and keep it relevant.
3. Once the survey is sketched out, use a word processor to create the questions and responses. That allows you to use a spelling and grammar check. Then cut and paste items into your survey program.
4. Right from the get go, clearly explain the reason for the survey, and how it will benefit participants if they take it. Make it clear they will (or won't) see the results, and that the material is kept private and confidential (or not).
5. Keep it short. Warn people how many questions are in the survey right away--in the introduction or the e-mail inviting them to take it. Knowing it's short will encourage them to participate.
6. Make the survey uncomplicated. It's very easy to get carried away with lengthy questions, cute answers, and convoluted matrixes and grids. Don't. For every degree of complication, you lose a bunch of people. I say this from personal experience: Check out question 14 on my survey, which I thought was funny. It bombed.
7. The paid version of SurveyMonkey allows you to force the survey taker to answer questions without skipping any. I discovered that some users quickly get testy when faced with that (it's probably an authority issue they picked up as kids) and I recommend you use it judiciously.
8. Include an incentive for taking the survey. I always offer an all-expenses-paid vacation in Hawaii or a free car; you may want to try something less expensive.
9. Use HTML sparingly and mostly for links in instructions. If you're HTML challenged, go to Webmonkey (no, no relation) for a tutorial.
10. It's critical that you test the survey to make sure it's error free, it's structurally sound, and makes sense overall. I always test a survey on a handful of trusted friends for feedback before making it public.

Here are three sites to visit for more details: "How to Design a Survey," "More Tips on Web Survey Design," and Perseus Survey Tips.

Sign up to have Steve Bass's Home Office Newsletter e-mailed to you each week.