Although I am a new faculty member, this year I have interviewed 14 candidates for 3 positions in 2 departments. Interviews at Prodigal U take 2 days. On both days, most of the time is reserved for 1-on-1 or small group discussions. On day 1, the candidate gives an open seminar to the department. On day 2, the candidate gives a closed talk/has a detailed discussion about their research plans. This depends on how the search is being run. In one department, the search committee makes a recommendation, but the whole faculty discusses each candidate and votes on whether to make an offer. In this case, the research plan is a very informal talk given to the department members. In the other department, the search committee decides who to make an offer to, and the department is not directly involved, other than to submit comments. In addition, our candidates are usually hosted for 2 lunches and 2 dinners as well.
As a faculty member, and member of two search committees, I can say that searching is very time consuming for the department. Everyone takes this very seriously, since making a bad choice can potentially have repercussions for 30+ years. So what are we looking for?
By the interview stage, all of the candidates are well-qualified on paper. Once we invite someone to campus, we want them to wow us. We are already impressed with the candidates' accomplishments, now we want to see their polish. They have all had research success in the past, and all have some interesting ideas for the future.
The first thing we want to see is that the candidate "walks the walk, not just talks the talk". We all know people who have been carried along by a successful PI and/or research group who have great hands in the lab, but lack that creative spark required for a good research plan. Thus, it is really important that the candidate demonstrate their creativity/scientific thinking as well as their breadth and depth of knowledge. Giving a good seminar is critically important. However, it is also important that the candidate can discuss their work coherently, and not just in the context of their job talk. We give a fair bit of importance to how the candidate answers questions and to how they discuss science in the interview meetings.
In discussion of the planned research we want to see short, medium, and long term plans. These plans should be differentiated from the work done as a PhD student or postdoc. The candidate should have some idea of how big their ideal group would be, and also to know what they absolutely need equipment-wise to be successful. Better still is a list of big ticket items and an estimated cost. Especially for candidates coming from large and well-funded labs with access to unusual equipment, it is important to know what infrastructure the planned research requires. If Prodigal U doesn't have it, there is no point in coming here. Two questions that commonly get asked in these sessions are "who else is working in this area?" and "what is your angle that makes your work unique/different?"
In terms of fit, we want the candidates to be reasonably pleasant and polite (berating the departmental administrators is not a good idea), to have given some thought as to how their future work will fit in with current research in the department, and to refrain from outright sexist/racist/homophobic/anti-Semitic/etc. comments and behaviors (this should go without saying, but you would be surprised). We look at candidates energy level--are they excited by their work, or is it just something they do. We are looking for someone we wouldn't mind having in the office next door.
The fit thing cuts two ways. We want there to be enough overlap in research interests that the candidate can find collaborators in the department, but not so much that there is direct competition for students and funding. This can be a fine line. The optimal amount of overlap can vary from person to person. I enjoy collaborating with my colleagues, so I am fairly tolerant of research overlap. One of my colleagues is more territorial, and would prefer to have almost none. At the same time, we don't want to hire someone whose research is very far from everyone else in the department, as lack of sharable equipment and difficulty attracting students can be fatal to a new lab.
Of the 14 candidates I saw, 5 performed "below the bar" and would not have been hired under any circumstances. I've been told this is not atypical. For 3 of them, their research talk was awful (hey maybe that was me in year 1 of my search!) The other 2 did not have convincing or realistic research plans. Another 1 who was above the bar prompted questions because the proposed research was incredibly similar to work done in their PhD and postdoctoral groups.
For the remaining candidates, there is nothing they did wrong. It is just "right place/right time" luck that gets them the job or not. Of the other 8 candidates, 3 accepted offers elsewhere and 2 accepted offers here. The remaining 3 candidates were ranked below the 2 that we hired. 1 search went unfilled (and will be redone next year).