When giving tasks to participants in usability sessions:
DON'T: "Tell me what you see on this dashboard"
DO: "Check if your e-commerce site is going ok."
Looking at a dashboard with no particular intention is a very different behavior than looking at a dashboard in order to figure out if the e-commerce site is still up and running.
User researchers should gather the usability issues that come up during realistic user behavior. If we are collecting data about fantasy behavior that is not helping anyone. That's just invalid data.
Realistic user behavior starts with a realistic user task. "Look around and tell me what you see..." is not such. There aren't many apps that people open just because they wanna read some UI copy and see boxes and arrows.
Here is another example to make the point:
Let's say a UXR at IKEA wanted to test the usability of an assembly manual.
- Researcher: "Look at this assembly manual and tell me what it shows you."
- Participant: [Reads the manual carefully, from point to point linearly. Doesn't skip anything. Spends a good 15 minutes reading... Has a little problem with reading the small print.]
- Researcher: "Assemble this bed. Here is the full package [assembly manual, parts, tools]. Start now."
- Participant [Not looking at the manual. Tries putting parts together... Gets stuck. Skips to the 6th page where she recognizes some parts... Goes back to the first page to disambiguate the parts... Realizes she put together the wrong parts... or did she?... going back and forth...]
Two completely different behavior with very different usability issues. One of them is an unrealistic behavior that was prompted by an exploration task. The other is realistic behavior prompted by a good task.
Only do exploration tasks if that's something real users would do. The example that comes to my mind is landing pages. Users visit those pages specifically to explore the product, and to understand whatever the page has to say.