“Tiff’s Tips for Developing Web Surveys”

I finally finished developing phase one of a survey at my employer. It was my first large-scale development project (large-scale in terms of the number of people we were asking to take the survey — 15,000), and the first time I had to develop such a project.

Because of time constraints, and my own lack of experience, this was not a snafu free project. But I learned some valuable lessons that will help me — and maybe you too — for the next go-round.

With that I present “Tiff’s Tips for Developing Web Surveys (And Other Stuff).” Couldn’t think of a snazzier title. Apologies.

• Get all of the requirements up front. Not just the high-level, “this-is-what-it-needs-to-do” stuff. I mean everything. A couple of the snafus in this survey came about because I had to make a change in one place and neglected to change the code or database structure in another. For example, for which questions will you need to have text input, versus check boxes, versus radio buttons? You’re much less likely to make a mistake if you know what you’re trying to achieve before you start coding.
• Think of the logic first. This is sort of related to the first point. Instead of thinking of the code first, think of the logic. What is it supposed to do? What if someone answers this way versus another? Nail down the logical flow FIRST, then figure out how you’ll code it.
• Audit your code before going live. When you’re in the middle of coding, it can be hard to see a mistake. There were a couple of errors I easily spotted on review that could have potentially been caught and fixed before going live.
• When developing, show error messages. In hindsight, it sounds like common sense, does’ it? Doing so just makes life easier all the way around, and can help you catch errors before going live.
• Go ‘sorta-live’ first. This is basically the same thing as beta testing. But having a lot of results (a couple of hundred is fine) can help you see emerging patterns more quickly. Use this as a first round so that you don’t find out later — as I did — that you didn’t some critical capture data.

Lessons learned for next time. Look for a how-to later this year.