VOC is not an excuse to abandon all other forms of collection. In many cases, VOC surveys ask visitors questions that are unnecessary, because the answers can be found elsewhere. This is a sign that an organization’s web monitoring tools are siloed: the people who know about performance aren’t talking to the folks who run analytics, who in turn aren’t sharing data with the people in usability.

Unnecessary questions reduce survey completion rates because the longer a VOC survey is, the more likely people are to abandon it rather than giving you the insights you need. So if you can get an answer somewhere else, don’t waste your respondents’ time—they’ll only give you three to five minutes of it.

Perhaps the biggest criticism leveled at visitor surveys is their sampling bias. It’s true that only a certain kind of visitor will respond to a survey, however good the invitation. While larger samples can mitigate sampling error, the answers you get still won’t be representative of your user base. You’re less likely to get responses from power users who are in a hurry, and even then, they’re probably only going to offer feedback in certain situations. And you’re more likely to hear from zealots and outliers.

This criticism misses the point: VOC should capture insights that you may be able to investigate. In the travel site case mentioned above, a few responses saying that visitors were just checking availability prompted the site operators to research further and confirm that this was, in fact, the case for many of their customers. Then, through analytics and experimentation, they were able to adapt their sites.

The best way to understand the needs, emotions, and aspirations of your target market is to visit it where it lives—in Facebook groups, chat rooms, Twitter feeds, news aggregators, and blog comment threads. You can use VOC to find out where your market hangs out, or to dig deeper into something you hear online, but you need to marinate in your community to really understand it.

If you need to constantly poll your market to understand its needs, you should convince visitors to let you contact them through email or RSS feeds. This lets you go back to them several times with additional questions and build your own panel of respondents. Remember, however, that enrolled respondents—and friends you interact with on social networks—are more loyal and “tainted” with opinions. After all, they liked you enough to enroll. So while it’s good to survey them, you still need to examine newcomers.

In other words, VOC works best on visitors who you don’t yet know, and who are new to the site, as soon as they’ve formed an opinion of you. Ask them too soon, and they won’t have visited your site; ask them too late, and they may have left forever.