Taking Google Surveys for a Test Drive
Six months ago Google launched an intriguing new way to conduct cost-effective surveys that offers an alternative to omnibus surveys. One or two easy survey questions are presented to online users as they seek access to high quality media sites. They gain free access in exchange for answering the survey questions. Google tracks how many people are answering each question and manages the process to ensure that the sample of respondents answering each question closely matches the overall U.S. population (based on Census data for those who have Internet access).
Here are some of the more interesting aspects of their approach:
- You can ask multiple questions, but the questions are answered by different samples for each. There is no way to cross tabulate or correlate answers to different questions you ask.
- Respondent burden is low, exceptionally so. Most people enjoy answering survey questions, but not zillions of boring ones. Here the burden is a mere fifteen seconds of their time.
- Demographics are inferred from online behavior using tracking techniques developed for targeted advertising. They infer gender, age, geography, and income, and claim that the data are 80% to 90% accurate. (Hmmm, maybe good enough for advertising, but is it good enough for survey research?)
- The cost is phenomenally low even compared to inexpensive omnibus surveys. The cost for a single question, fielded to a representative sample of 1,000 U.S. adults is $100, compared to at least $800 for entry-level omnibus surveys.
- Turnaround is fast, in most cases just one and a half days. The fastest omnibus surveys typically take three to five days.
It is worth considering this tool if your research needs are exceptionally simple. How simple? Well, we tried to implement a basic omnibus project using this tool, but it wasn’t possible. We needed some branching logic and also needed respondents to compare three brands in a matrix. Google Surveys does not do this.
So if you have one or two super duper simple questions and you just need a quick read on what people think, give it a try. Anything more, and you’ll need to consider alternatives such as traditional omnibus surveys or a custom survey that is carefully designed to get exactly the answers you need.
–Joe Hopper, Ph.D.