The most important aspect of being able later do a comparison with the data, is to receive data that is totally objective.
This means you must ask the poll takers specific objective questions, perhaps even only allowing them to choose from a bullet list of objective answers.
If you allow freeform answers, you most likely to get subjective / emotional answers, that boil down to political pie charts.
ie, 60% of users love rendering inside SketchUp, whilst 40% hate it, and would rather render in an stand alone application.
That kind of emotional data really doesn’t help someone new to rendering choose a renderer.
Example. Do not ask a user if they “like” or “need” a certain kind of feature. (That will get a subjective answer.) Instead ask them specifically how often they use that kind of feature. (With choices like: “Every time I render”, “Most times I render”, “Occasionally when I render”, “Seldom when I render”, “Never”, etc.)
The idea is to get data on what people really, truly need and use for features and settings.
Which often turns out to be different than what they think (or think they remember) they like or love about their renderer (emotion again.) Added to that, is the sneaky marketing that tugs on people’s “heart strings” and convinces them to become emotionally attached to products.
… my 2 cents worth.
(…and if you click the little heart icon below this post, It’ll be a subjective thing you do.)