Related topics: Collected Wisdom, doing science

Collected Wisdom on Job Talks – Building a Truly Great Research Seminar (UPDATED)


This is the second installment in the Collected Wisdom on Job Talks series. The first

1) Provide a narrative: the art of clarity and mystery

Start big – Good talks begin with a topic that everyone in the audience can understand and relate to. The goal of the first few slides should be to capture the audience’s imagination and make them feel invested in the story you are about to tell. In other words, create common ground before diving in.


  • Care about story telling and narrative structure. The best seminars are good stories and like all good narratives they have an interesting and engaging plot. They keep the listener at the edge of their seat. The art of a good narrative is to combine predictability and surprise. Tell your audience what the story will be about (this is what “organizing slides” are about) but have a plot with surprises and interesting turns. The art of a good narrative comes from either being an innately good story-teller, or from reading good fiction (Poe, Hemingway, some of the good pulp and mystery writers such as Elmore Leonard are great models). It does not hurt to make an outline of your plot rather than just putting a bunch of slides together. Posing mysteries/questions is a ploy that can be used to great dramatic effect.
  • Use a story-telling technique other than chronological – “first I did this, then I did this, then I did that” is, well, boring. Extra points for creating a suspenseful story. Two story-lines that work really well:
    • 2×4 from the left – leading your audience in one direction, with the implicit expectation of a certain result or outcome, dashing their expectations (the 2×4), using that research “disaster” to realize some greater pattern, outcome or underlying fundamental truth (this structure works well for a series of experiments with early “failures” leading to a divergent path and final success)
    • Baptist preacher – start high, bring low, end high. Works really well for conservation talks where the end message mixes scientific outcomes with a larger social/societal message of interaction/responsibility/ etc.
    • Science is a detective story. Tell it like one. State hypotheses explicitly, with at least two strong, plausible alternatives so that nobody can figure out in advance what your results will be. Highlight the unexpected and the counterintuitive results. Play up any clever insights that let you solve the problem, or any exceptionally large datasets and analytical savvy that let you do what nobody else has been able to do before. Science is (or at least should be) a CREATIVE enterprise, not just a slog.

2) Provide context and structure

  • Care about didactics. When you give a lecture you are both an entertainer and a teacher. Audience members should go out knowing something new and having learned how to explain it in a couple of pithy sentences. The best talks are those in which people leave saying “I didn’t know that! Man, feather lice are so cool!” Do not be embarrassed about giving all the background/context needed for the talk to be understood and for its importance to be appreciated. What is obvious to you might not be to the rest of the audience. Ecologists will welcome a refresher on how G proteins work and physiologists will thank you if you explain how the neutral theory works and why it matters.
  • Why is your research important to science? What are the big questions you’re addressing? Why is your system BY FAR the best way to get at these questions? Context and relevance are EVERYTHING for the majority of the audience.
  • Have waypoints with clear summaries — building to a grand conclusions
  • Create an interesting storyline and communicate the organizational structure that goes along with this storyline to the audience. Good talks let the audience know where they are headed and check in with them along the way to keep them posted on where they are at in the storyline.
  • Start with saying what you are going to say. It could be an outline of the talk or even some of the major conclusions right up front. People want to know within less than 5 min what the point is going to be and whether they are in the right lecture. Only after this brief statement, begin an introduction of the general area of this work and why it is interesting. After an introduction, you have your observations, and towards the end, summarize all over again. Include broader implications. Don’t forget this last section on implications, else people may go away wondering, “so what?”
  • Have a clear overarching question that you start with, flesh out, and return to (i.e. say what you will say, say it, remind people what you said). The bricks are the excellent results and data you have garnered, but without the mortar, no one will understand their significance. Outlines, schematics with parts highlighted, etc, are great for this.

3) Keep it simple

  • Have a good conceptual framework explaining what you like in Biology (or whatever field…). Usually only a small fraction of the audience will really get the details of what you do. They need to hear and see (with a schematic) what the big picture is. This is usually three topics, like “behavior, physiology, and environment” or “modeling, field observations, and manipulations”. It really can be anything…but often this is the ONLY thing that the majority of the audience will take away about the candidate. It gives people an idea of what you would do for 30 years, and how you would fit into that department.
  • Keep the main points simple – People probably only remember 1-3 things from your talk. Make sure that you decide which 1-3 things are the most important and that your entire talk revolves around them. Remind your audience of those main points at the end of your talk as well.
  • Bad seminars frequently try to pack far too much into the time slot. My gut tells me that you can hit only 2-3 pieces of the big picture story in any seminar.
  • In any talk you will have some goal–a few points to make. Organize the talk to maximize the clarity of these points. Throughout focus on items that lead you to these specific goals and reject tempting tidbits that don’t aid in getting there.

4) Care deeply about representation: from graphics, to art to font

  •  Slides must be clear but in great seminars they are beautiful. Spending time making beautiful images that are didactical and clear separates the good/competent talks from the excellent ones. What we do, the creatures and systems that we study, are so damn beautiful that there is no justification for esthetically sterile talks.
  • Use an artist’s eye when making your slides – Use a combination of background and text colors that contrast with each other but do not make your eyes strain. Try to place objects on a page so that they are evenly spaced and “look nice”. In some cases using a photographer’s ”rule of thirds” can make slides look very professional. I like to outline all of my photographs and figures with a thin border so that the images look crisp. There are lots of ways to make your slides enjoyable for an audience to look at. I have gotten many comments on this over the years and it can make a real difference in keeping an audience’s attention.
  • No slides where ANYTHING is unreadable or hard to interpret. How many times have you heard a slide introduced like this: ”I know you can’t read this in the back of the room, but …” ”These colors didn’t show up as well as I thought, but …” ””This is a very busy figure so I’ll walk you through it …” ”I know this slide is a little dark, but trust me — this little smudge is really important …” ”There are no labels on these axes, but they are …” IF YOU HAVE TO APOLOGIZE FOR A SLIDE IT IS NOT WORTH SHOWING.
  • Clear and beautiful figures that are clearly explained. What is the x-axis? What is the y axis? What pattern (trend line, colors, differences, no relationship, etc) do you want people to see in the figure? Only after explaining all that will people really understand the significance (which you should also tell them).

5) Cut text, then cut more text, then remove even more text.

  • No text-only slides. EVER. Images should be compelling, clear, and large enough for the old timers in the audience.
  • Use PowerPoint as an image projector, not a presentation crutch. That means minimal text, rarely if ever in bullet point form, always large font.
  • Use as few words as possible. Avoid long lists of anything. People can’t retain them. If you have to say, “I know you can’t read this but….,” you have too much material on your slide.
  • Make sure that you’ll talk about everything on each slide. Otherwise, remove unneeded figures or panels.
  • Minimize text on slides – put only enough text on the slide to remind yourself of what you need to say, or very very strategically to remind the audience of some key concept or number that they need to remember as the story rolls out[1]. And if you can remind yourself what you need to say with just a photograph or a graph, even better. This produces a more professional slide set that seems a little more ‘mature’, and it draw the audience into the speaker’s story rather than reading slides.

6) Know and care about your audience

  • Know your audience. The first 10 minutes and last 10 minutes (at least) must be accessible to EVERYONE in the audience. I have never, EVER heard anyone complain that a speaker spent too much time on introducing the subject and putting it in context.
  • Know your audience. A seminar to a biology department is different than a seminar to a cell biology or ecology/evolution department.
  • Know the audience. It’s crucial to gauge the level appropriately. If you know key people in the field will be in the audience, make sure that you cite their work appropriately, but not gratuitously. Also, don’t confuse appealing to a broad audience with “dumbing down” – colleagues (including undergraduates through senior faculty) from other disciplines or sub-disciplines are not stupid, even if they aren’t up on the latest bayesian technique,
  • For the job seminar, talk briefly in a very genuine way about your fit to the department. This is not an idle shout out to all the professors you might have something in common with. Instead, this takes researching the departmental resources and understanding how you could both use and contribute to them.
  • For the job seminar, use the last 5-10 min. to talk about what you would do in your new lab. This could even have potential specific aims for a grant. Consider what experiments your first rotation or grad student or undergrad(s) would do in the lab.

7) Be clear about YOUR research

  • Be very clear what YOU, PERSONALLY have contributed to the science. Own the work. Don’t say “I’m a postdoc in the xx lab and we study….”. This is especially important if you’re coming out of a big lab. Some tips:
    • Use your acknowledgements well. Be classy, complete, and quick. This is one of the only times in your seminar when you will put text and information on a slide that you will not talk about. List funding, list advisors and lab-mates, list collaborators. Don’t call them all out by name.
    • List published works under relevant results, so folks know what is published.

8) Reduce the number of slides and the number of transitions

  • The exact number depends a lot on how you use them, but you will almost always have too many
  • Up to 35 slides for a 45-minute talk might be possible. More than that, especially if many have data, is just overload. For a 15 min talk, 10 slides seems the upper limit. You may be able to get through more, but the audience will not. You want them to come away satisfied rather than frustrated. Boil down the message to the essentials.
  • Avoid data diarrhea: Sometimes your story has lots of data you are very proud of. One approach is to show the audience how you carefully analyze one or two points using the raw data. Do this until the audience recognizes that you think well and trusts that you are satisfying rigorous criteria. Then say that you investigated the five other points with similar methods, care, and scrutiny, but since there isn’t time to show each of the individual experiments, you will just be stating the results. This way, you have illustrated how to do it well with details and yet not dragged the audience through too many details.
  • Keep slides simple – Do not use a lot of words and do use large and simple photos and figures. I try to make a single figure take-up the whole slide so that it is easily visible and I can talk through the key points rather than list them with words. Similarly, I use photographs on slides while going over discussion points rather than listing them as bullets.

9) The E’s of life, in 15-minute packets, and with clarity

  • The art of giving a great talk boils down to 4E: Empathy, Ethics, Esthetics, and Etiquette
  • Any talk over 15 minutes is longer than our brain’s attention span, so break up your talk into 15 minute pieces and then pause for a minute of so between each one so that people can rest.
  • Don’t try to show your disciplinary colleagues how well you speak the jargon and stop trying to “out-math” them – live in fear of a resurgence of buzzword bingo, and strive to make your work understandable and relevant to the broadest audience


[1] Beware: This latter use of text is where speakers go horribly wrong, thinking the audience needs to remember dozens of full sentences and big bulleted lists.

UPDATE: 10) Give the audience some meat

Jeremy Fox added this important point: When thinking about the scope of your talk, don’t be too superficial. Yes, you need to give the audience the big picture of your science rather than just narrowly focusing on one specific project. But you also need to “give the audience some meat” (as Jeremy’s colleague once put it). You shouldn’t just skim over too many projects, that doesn’t allow your audience to appreciate and evaluate your science. Rough rule of thumb: go into at least some depth on 2-3 projects or lines of research.

Recent Posts

Monthly archives