Avoid Do It Yourself Qualitative Research

More than with any other analytical tool, there is something about qualitative research that makes everyone an instant expert, and this is not necessarily a good thing. The design of qualitative research projects is a specialized skill that is learnt and honed, just as the ability to formulate a targeting model is a specialized skill; the fact that the former does not normally resort to mathematics does not make it less analytical in nature.
Ideally, an analyst would be given a business problem and a budget with which to design a study. He would then decide on the best type of study, the scope, the sections of the study as well as the details of the execution. There should be some close dialog with the other stakeholders to verify that the all the requirements are being addressed, but - in this ideal universe - the non-analytical stakeholders provide only inputs (i.e. the formulation of the problem) and evaluate outputs (i.e. does the study answer the questions that I want answered and does it raise further, useful questions?) but refrain from dabbling in actually designing the research study.
Unfortunately, however, there is the pervasive view that qualitative research is mostly "a matter of common sense", and that everyone is qualified to design it and, to some extreme, to execute it. The risks of this approach are:

• biased results - a poorly constructed interview or focus group may lead the respondent to answer what we expect her to answer, and frequently confounds aided recall with unaided
• compromised creativity - too rigid a discussion plan keeps participants from straying off into some creative perspectives about the problem being researched
• creativity overdose - this is the exact opposite of the previous case: in some cases it is preferable to tether the conversation to a line of inquiry that discourages respondents from wandering off topic
• non-sequitur - section B does not follow from section A and section C does not follow from section B, creating smooth transitions within a discussion is an art that takes some time to develop, when transitions are brusque, respondents tend to require some "adjustment time" that may or may not have been accounted for in planning the length of the session, I have seen cases where the transition was abrupt enough that the participants had to be briefed about the details of the new topic being discussed. In an extreme case a non-sequitur comes from trying to investigate more than one topic in the same session... bad idea.

The list of problems that can result from poorly planned and executed qualitative research is longer than shown here, the solution is simple: let your analysts know what you need to know from the market, outline the business questions you want the research to answer, and then let the analyst use her experience and knowledge to design a solid research plan.