If you did get that interview invitation, your file clearly looks interesting or good enough to the committee; now you need to undergo a different kind of selection process.
Your audience just got bigger: people marginal to your interests now meet with you or try to find out if they support your application. Any talks you give about your work must be broadly understandable. Understand how broad your audience will be! Perhaps not people working on insects, not on behavior. PRACTICE your talk and elevator speech with similar non-experts. You must find ways to relate your work to pretty distant areas. Is your work on insect behavior relevant to robotics, to community ecology, to fish, to evolution of multicellularity, to engineering, to evolutionary biology, etc.? Discuss this with your advisor and other mentors.
- Applying for grad school: you likely won't be asked to give a seminar talk, but you will be asked by different people over and over what resarch you've done, what your future research interests are. You must have smart answers prepared for each of these questions, of varying lengths depending on how interested the person you are talking to actually is. This is also called the 'elevator pitch'. Check Joan Strassman's interpretation of this.
- Applying for a postdoc position: you're probably just going to be interviewed by your prospective advisor, perhaps a collaborator or two, or her lab members. But quite often you will be asked to give a talk, and other department members may be there.
- Applying for a faculty position: you will be asked to give a talk to at least one department, perhaps to an interdisciplinary audience. For this it is crucial that you understand how broad your audience is. Add photos, analogies, explain every acronym, every method. This is the main point where many interviewing candidates fail: they fail to generate interest and excitement in people who don't work in their area. Your seminar talk is the place where you must do this.
Personal likeability: Everyone (!) you meet will potentially be asked to judge you. Most of these people will not read your file, and even if they have, they will often not remember or care much about it. Interviews are the time to get a 'feeling' for the candidate. This is obviously fraught with subjectivity, perhaps even self-delusion (and there are studies on this). What follows are my impressions on what makes a candidate leave the best impression.
Avoid cultural mismatch: If you are European (like me), or any nationality other than where you are interviewing, seriously think about and study different cultural habits. Europeans in the US must usually be more open, more enthusiastic, less cynical than what their instincts tell them (do not in any way qualify=narrow your level of enthusiasm for the position, or the relevance of your work). Also try to avoid a strong accent. Practice and over-practice your talk/elevator pitch so it is fluent and clear.
Positivity: Do not be in any way ambiguous or negative about academia, fundamental science, the place you are interviewing, etc. In fact try to avoid any kinds of negative statements, even about weather, food, or whatever. Be positive all around: simply emphasize the positive or choose to phrase your sentences positively. People like positive people more than complainers. Care about, i.e. be interested in and enthusiastic about all research, the natural habitat, the place and people you encounter.
- Applying for grad school: You are likely interviewing with faculty and with grad students. Both should come to like you. But in talking to faculty, you should concentrate on impressing them with your maturity and research expertise. In talking to students, you should focus on finding out more about the program. However: Do not ask negative questions, do not force people to delve into problems. It will leave them with a bad taste that they'll associate with you. Read between the lines instead, and you can always ask more probing questions once you've been accepted (by email).
- Applying for faculty positions: Good, active departments want to hire great, active faculty, but they also want to imagine that the new faculty will bring reknown, grant money, and fun to the department. Don't underestimate the last part: faculty want new faculty to pull them out of their rut, to allow them to see the world as shining and new again, to be again on fire with the excitement of discovery. You as interviewing faculty candidate have to be like that yourself and also be able to light the fire in others. If you do this, or if you are known as a fun, interactive, creative, broadly interested and educated person, this will substantially increase your chances. This is no matter what they are actually looking for, and no matter whether individual faculty are likely to collaborate with you.
'Fit': Everyone will proclaim that no matter how good your file or how nice you are as a person, what they care about is whether you 'fit' into a particular lab, program, or department. This may or may not be another version of assessing likeability. If there really is something specific that your potential advisor/department is looking for, you have little influence on this. However, it cannot be overstressed how much love begets love: if you are lukewarm about the position, the people there will be lukewarm about you. If you clearly love the position, the people there will have a hard time not loving you. This means:
Commitment: If you are interviewing for your top choice, say it. Say it afterwards if you realize it afterwards. Sending emails to people you met afterwards is good form as long as you don’t spam: if you send one to every single person you met at every single interview this is going to be obvious. But do send an email to people you genuinely connected to, to your potential future advisor, to search committee members at places you really loved. In that email, specify what you liked about the place/opportunity. If it is your top choice say it.
Find loyal advocates: Very related to the last point. You must identify the advisor you plan to work with, or potential collaborators of different likelihood. You can exaggerate how big the second group is, but you must narrow and strongly identify the first group. Almost any position you may interview for, you depend vitally on someone in the department or on the committee choosing you as their favorite and arguing strongly for you. This is also true at grant panels. So if you know who this person is likely to be, you MUST convince them to put you at the top of their list. Meet with them as long as you can. Send a thank you email, and quickly (within a day). Tell them how much you loved being there, their work. Point out clearly and specifically what links you see between your research and theirs. If you are interviewing for a faculty position, this is less exclusive, but finding a group of 2-4 potential collaborators will be very helpful (and talk to them about the specific ways in which you see potential interactions – these don’t have to be about your past work, they could be what you envision a future jointly advised student may work on, etc.). But you can also excite people who have nothing to do with your field about your work and personality (see above).
More on giving that job-talk (i.e. seminar talk when applying for faculty positions, rarely for postdoc positions):
You should be honest about the limitations of your conclusions, but not too negative. Have strategies prepared for how to address shortcomings, not excuses.
Remind the audience of highlights of your file. Clearly identify which publications are yours, in which journals (never leave this ambiguous on the slide). All you are a coauthor on can be claimed by you, although you certainly acknowledge collaborators. Perhaps there are other highlights: your paper received media attention, you changed a paradigm in the field, you got funding from an unusual or competitive source: don’t dwell too long on this to not seem arrogant, but mention these things in passing. Do not assume any of it is obvious. *Do not assume anyone remembers your file*, not even the committee (or your prospective advisor if interviewing for postdoc).