A free comprehensive guide for evaluating site structures

Page tree
Skip to end of metadata
Go to start of metadata


As mentioned earlier, we may want to control who gets to do our study (and who doesn’t).

The easiest case is that we’re happy to get results from anyone who visits the site (in the case of web ads) or anyone we’ve emailed (presuming that being on the email list is enough qualification in itself).

However, there will be times when we want to be picky – when we a particular subset of users (or don’t want them).

For example, if we’re testing a site structure for a bank’s online banking service, we probably want online-banking users. More specifically, we may want to exclude bank customers who have never used online banking and never intend to, since their answers would be irrelevant and could actually skew the results.


Screening by filtering a database

If we’re recruiting using a customer database, we may be able to filter the users down to those who have logged into online banking, say, once a month or more. That may be enough “screening” for us, so we can go ahead and run our study without more qualifying questions.


Screening using separate email lists

If we have email lists that are already separated into groups that meet our specifications (e.g. business customers vs. consumers), then the job is easy – we invite people from the appropriate list to the corresponding study.

However, because email invitations can get forwarded (beyond our control), we may still want to include a screening survey question in the tree test, just to make sure we’re getting the participants we intended.


Screening with explicit questions

But what if we’re using a web ad (which every site visitor sees) or we have a customer email list that doesn’t include enough data to filter the users the way we want? In these cases, we’ll need to explicitly ask each interested person if they meet our specifications.

There are several ways to do this screening:

  • Using descriptive links
    The easiest way to point specific types of participants to specific tests is to use a list of links that describe who they’re for. We can use this list in both an email invitation and in a web ad’s explanation page. This works well as long as the descriptions are clear and distinguishable from each other:

  • Using a screening tool
    We can also use a dedicated screening tool (such as Ethnio) to make sure we’re getting the right participants. When users click our web ad, Facebook post, or Twitter tweet to participate, these tools pop up a window that asks the qualifying questions that we’ve provided. People who “pass” are directed through to our tree test, while the others are politely thanked and dismissed.

  • Using the tree-testing tool
    If our tree-testing tool offers a screening feature, we could of course use that instead.
    Even if it doesn’t, as a last resort, we could ask a screening question during the tree test itself (as a survey question either before or after the tasks), and then only include the qualified participants in our results. However, that would be a waste of time, effort, and data for those who didn’t meet our screening criteria, and we’ll need to include them in any reward we’re offering, because they did the study as we requested (even though we didn’t use their results).


Next: Restricting access with a password


  • No labels
Write a comment…