I am doing usability test on a product which I think might have problems with it's search function. I want to test search function with users, but I don't want to restrict user information retrieval methods by adding straight up "use search function" task. But if I make just general information retrieval task, it might end up in a way that hardly anyone uses search. Is there way around this problem?
If you can't isolate the variables within one test, the only real solution is to split it into two tests.
This might be mitigated by doing the user information retrieval test first and seeing if any search feedback naturally falls out from that. If it does, great. Analyze the data and determine if you need more specific testing. If feedback doesn't come from the first test, then you definitely need to move on and conduct the search-specific test.
I might consider a pre-test survey to collect the participant's predisposition to using search. Something like:
Q. When trying to find information on a website, I commonly use the site's search:
The question above needs phrasing work, but the idea is to get a user gauge on wether or not they're predisposed to searching for content on a website. Then you will know how much you will need to direct them to the task.
Other option is to move towards testing the participant's ability to complete a task. You may want to restate your tasks to: 1) Locating the search function 2) Validating search results compared with participant expectations 3) Filters and advanced search options, etc.
This way the test skews away from "If a member would use the search".