Quick Feedback function

If you want to “try before you buy”, i.e. test a survey and see potential research outputs ahead of launching it with your target audience, you can use Quick Feedback to assess your survey. For a small price of $20, you can get ~50 survey-takers (Amazon Mechanical Turk “workers”) to give you feedback within approximately 6 hours.

What’s included?

  1. Example research outputs based on the settings of your experiment
  2. Testers’ suggestions for:
    • The wording of questions, features, attributes, product descriptions
    • Presentation of items (for instance, alert you when the survey is too text-heavy and requires the use of imagery)
    • Informing you on how to make your survey more enjoyable so that your real respondents stay engaged
  3. Testers’ rating of the survey in terms of “Enjoyment”, “Recommendability”, and “Understandability”. Below is an example output of workers’ rating of one survey:
Overall feedback scores from Amazon MTurk survey respondents

Testers will also comment on which part of the survey is “hard to understand” and “what needs improvement”. Below is an example output of the open-ended feedback:

Example output of open-ended feedback from Amazon MTurk survey respondents

How to start it?

In order get your survey tested, select the Get Quick Feedback option from Other actions in your experiment settings:

Selecting the Quick Feedback option to add survey respondents from Amazon MTurk

Important caveats

  • Amazon Mechanical Turks are not representative of most target audiences.
  • They are not guaranteed to be located in any particular country or to use any particular language (but most will know English).
  • Some (or in rare occasions, all) feedbackers will be marked as “low-quality responses”, which will give you an indication of the level of engagement your survey will evoke.
  • You should only use this service for “quick-and-dirty” feedback, and not rely on any findings in the tests.
  • Please do not ask for any personally identifiable information, such as first or last names, emails, addresses, and phone numbers.
  • If you add any screening questions or logic to your survey, the number of complete responses will probably be lower than 50.