Having watched decionmakers ask questions directly to customers, I can confidently say that the skills and capabilities of the researcher in interpreting interview responses are almost always vastly superior to the non-researcher.

In other words, you're probably crap at it.

(Well, not specifically Roger Martin, I'd hope, but I've lost count of the times I've kicked a colleague under the table for saying 'would you buy this?')

The researcher asks the questions, the decisionmaker observes. You're not playing telephone, you're still there, but you're not screwing up the interview with leading questions. You can ask your own questions at the end, but if they're crap, I'll flag them in big red pen in the writeup.

"Are the questions really that bad?"

"To quote: 'how far would you agree that this product is critical to your job?'"

"I ... I see what you mean."