Hi Fairy, great question! Speaking from my own experience, it’s quite tricky to start by aiming for low-fidelity prototypes, especially if the testers are the average person / common user. The way these users interpret the low-fi prototype in front of them could be very different from the designer’s original intent — hence the need to make the prototype as realistic as possible. However this doesn’t mean spending two weeks building a prototype. It simply means that we should move fast in each iteration while trying to get to as realistic of a representation of the product as we can, without settling for something that doesn’t feel real.

This is not to say that low-fidelity testing isn’t ever effective. In fact it works very well with domain experts and power users. However domain experts are much more expensive and harder to source, and power users don’t really exist if your product is brand new!

Side note, regarding the topic of superficial feedback. Yes, it will happen. As designers and researchers, we need to be cognizant of that and use our best judgement to nudge the testers (without leading them) back to the critical issues during the session, and then use our judgement again in the debrief and analysis afterwards to exclude the feedback we aren’t looking for.

A lot of the work actually happens after the test. Hope that helps!