Usually you must ask about socieconomic aspects and other questions to characterize the profile of the informants. Then, you can ask about the use of plants and how this use is made, the aim of use of plants, and to improve your study you can collect a sample of the plants used by informants and do a botanical identification to check the correct botanical name and to make a link to the popular names. Some journals offer models of this kind of studies, such as Journal of Ethnopharmacology, Journal of Ethnobiology and Ethnomedicine, and others.
Depending on what is the purpose of the survey. However, whether the purpose is to choose plants to find compounds with biological activity, you could also ask about the organ of the plant used (roots, leaves, bark, etc) and also how they prepare the ¨"remedy" (infusions, decoctions, cataplasm...).
Its all depends upon nature of plant, agroclimatic conditions, demography, importance in daily routeen, nature of evolution and one can keep on adding, but one should asked about phermacutical properties indirectly by asking about its utility in healthcare and general upkeep.
There are no key questions. First you have to come down on the equal stage of the "traditional healer" otherwise you get not the right information. For this you have to fall silent. In your patient waiting and during your close-mouthed speaking with the healer you get all the main information you need first. Then you can try to ask about some plants and their use. I found that the kind of models for studying, which are publiced in some journals are only academic and do not show the bits and bobs you absolutely have to know for you survey.
Your question is extremely broad, but I actually just had a class session today about combined quantitative/qualitative ethnography design.
My advice is to read a pharmacological journal article profiling a drug which we don't know overly much about and replicate that empirical, objective, and extremely concise kind of descriptive language - but describe it in non-technical terms that anyone could pick up and read and understand (like the way Scientific American does with its articles).
Follow that up with an ethnography whose theme question is to understand what your research participant demographic's relationship is with the medicine in question, and rather than going in with a research thesis in mind - just start by asking lots of questions beginning with "HOW" (not "why", as the former word will get you more subjective, emotional, and humanistic data that a good ethnography needs.
Some example questions:
"How do you use this medicine?"
"Who uses it?"
"How often do people use it?"
"What do they use it for?"
"What does it do?"
"How does it affect people?"
"How does it affect your village/town/tribe/culture/society/city?"
"When do you use it?"
From there, you'll get a very clear descriptive picture of what that medicine means in that culture, and you can pick a focus for the paper itself at that point easily.