Focus groups tend to be one of the "go to" choices for early-phase consumer research, many times because the methodology is common, and resources, such as facilities and moderators, are generally easy to locate. I would argue that focus groups are not well suited to the needs of early-phase innovation research, the type which we would be prone to conduct.
I've spent thousands of hours both "behind the glass" and "in the room" in research facilities conducting fieldwork of various types, and have had the occasion to both observe and participate as a member in focus groups. Even in the well-moderated sessions, I tended to come away thinking about how skewed the discussions tended to become, and how much I would have wanted to interview the participants one-on-one.
The effect of the focus group format on validity of findings as opposed to one-on-one depth interviews have received quite a bit of scholarly attention, and papers have been written exploring methodological issues with focus groups. This abstract from Boateng (2012) [1] summarizes the findings of the overall body of research:
The efficacy of Focus Group Discussion as a qualitative data collection methodology is put on the line by empirically comparing and contrasting data from two FGD sessions and one-on-one interviews to ascertain the consistency in terms of data retrieved from respondents using these two data collection methodologies. The study is guided by the hypothesis that data obtained by FGD may be influenced by groupthink rather than individual respondents' perspectives. A critical scrutiny of the data that emanated from the two organized focus groups discussion departed quite significantly from the data that was elicited from the one-on-one qualitative interviews. The difference in responses confirms that FGDs are not fully insulated from the shackles of groupthink. It is recommended, among others, that though FGD can stand unilaterally as a research methodology for nonsensitive topics with no direct personal implications for respondents; researchers should be encouraged to adopt FGD in league with other methodologies in a form of triangulation or mixed methodological approach for a more quality data, bearing in mind the central role occupied by data in the scientific research process.
Furthermore, in my experiences, the group discussion format of focus groups tends to elicit the following behaviors, each of which has its own way of biasing or eroding the findings. (If the group is composed of 18-35 year-old males and females, multiply the biasing factor by 3):
My overall point with this example is that, much like any social gathering with people who do not know each other, when you get 8-12 people in a room in a single conversation, people "overact" or take on personality traits they otherwise would not. Impromptu caucuses will form before your eyes, as people with similar thoughts will band together.
I offer the following as a humorous example of some of the exhibited traits you might see in a focus group. Interestingly, the morning after this Saturday Night Live sketch aired, the consumer research world exploded in agreement and story as to how realistic "Linda" was! Please watch the following 6:49 video.
If we break apart some of the research "jobs" we would likely need in evaluating ideas or early-phase concepts, the role of the focus group becomes more and more niche.
Surveys do a better job at understanding the overall space than focus groups, and are far less expensive. Furthermore, you are receiving "clean data" in a survey, unbiased by social pressures of the group or groupthink.
Individual interviews will give you far more depth than any focus group, while allowing the interviewer to explore topics and ideas of interest.
Observation and ethnography will tell you more about use phase in real application. Individual interviews will tell you more about initial impressions of a prototype in a controlled environment, free from group biases.
Message or proposition testing in a live environment will provide far more realistic, specific, and practicable information.
This leaves us with focus groups being used as ideation sessions to generate ideas and creative. Needless to say, in these applications, focus groups may have far more in common with the SNL skit than you might like.