How to Build a Good Research Presentation

The last post in this series covered what I look for when I watch a talk: the standards, the traps, the things that quietly signal thin thinking. This one is the other half. Given those standards, what does a talk that meets them actually look like, slide by slide. Think of it as the outline I would hand you if you sat down in my office and asked, “okay, where do I even start.”

I will walk through the structure roughly in the order it appears on stage. For each part I will explain what it needs to do, how long it should take, and what a slide that does its job actually looks like.

The shape of the whole talk

Before the slides, the shape. A research talk is an argument, not a report, and an argument has a spine:

motivation, problem statement, gap, literature, theoretical grounding, research question, methodology, results, discussion, future work

Every section exists to make the next one necessary. Motivation makes the problem statement land. The problem statement makes the gap obvious. The gap makes the research question the only reasonable next move. If you can remove a section from your own outline without the next one losing its footing, that section was not pulling weight.

For a fifteen-minute talk, a common slot for a lab meeting or a qualifying exam, I roughly budget it like this:

Section	Time	Slides
Motivation	1.5 min	1
Problem statement	1 min	1
Research gap	1.5 min	1
Literature review	2.5 min	2–3
Theoretical framework	2 min	1–2
Research question(s)	1 min	1
Methodology	3 min	2–3
Results	3.5 min	2–3
Discussion	1.5 min	1
Future work	1 min	1

That is roughly nineteen minutes, which already tells you something useful: your first full draft will run long. Everyone’s does. Budget the cut before you build the deck, not after.

Opening: motivation and problem statement

Open with the problem, not with yourself, not with your field, not with a technology. The first slide’s job is to make someone who has never thought about your topic feel, within thirty seconds, that the problem is real and matters. A motivation slide should follow this shape:

Something the audience already recognizes as important is currently hard to do well. The usual ways of dealing with it fall short in a specific, describable way. That shortfall has a real consequence for real people. So a new approach is worth pursuing.

Then narrow it, on the very next slide, into the precise version of the problem you are solving. A problem statement should be one or two sentences, specific enough that someone could not confuse it with a neighboring problem in the same field. If your problem statement could describe five different projects, it is still too vague.

Notice the movement across those two slides: broad and human on the first, sharp and specific on the second. That narrowing is the whole job of the opening.

The gap, stated as a bridge

The gap slide is where the literature review earns its keep before you have even shown it. State the gap in one clean sentence before you build the case for it. The shape is simple:

Prior work has established X. It has not, however, addressed Y. That is the space this project occupies.

Everything in the literature review section that follows should build toward that sentence, not wander away from it.

Literature review: synthesize into two or three slides

Two to three slides, not ten. The instinct to show breadth by listing every paper you read is understandable and wrong. A talk is not a bibliography, it is an argument, and citation dumps do not argue anything. Organize by theme, not by paper.

A useful pattern is three slides in sequence: one slide summarizing what the field agrees on regarding one relevant thread of research, a second slide doing the same for a second relevant thread, and a third slide showing where those two threads have not yet met, which is exactly where your gap sentence lives. Each of the first two slides should synthesize four to six sources into two or three sentences, not list them one by one.

If you find yourself wanting a fourth or fifth slide “just to fit everything in,” that is usually a sign you have not yet decided which papers actually matter to your specific argument. Cut, don’t add.

Theoretical framework: your own, defensible diagram

If your work is grounded in a theory, whatever theory fits your field, this is where you show how it maps onto your problem. Build the diagram yourself. Every box should be a construct you can define without notes. Every arrow should represent a relationship you can justify, ideally one with a citation attached to it. A framework slide that survives the question “why does this arrow point that way” is a framework slide worth having.

Research question(s): the one slide everything else points at

State your research question or questions plainly, numbered if there is more than one, and phrase them so precisely that a stranger could design a study from them alone. A well-formed research question names the variables or constructs involved and the relationship you are testing between them, in language specific enough that two readers would design roughly the same study from it.

This slide is short on purpose. It is the hinge the whole talk turns on, and hinges do not need decoration.

Methodology: enough to be reproduced, not everything you did

The methodology section should let a competent peer understand what you actually did and, in principle, redo it. That means participants, materials, procedure, and measures, but at the level of a talk, not a methods chapter. Use a process diagram if the procedure has sequential steps. Use a table if you are comparing conditions. Use plain text only when neither adds clarity. One useful test: if a slide in this section is pure prose with no diagram or table and the procedure has more than two steps, ask whether a visual would actually help before defaulting to bullets.

Results: show the finding, then say the finding

Every results slide should pair a specific figure or table with a one-line, plain-language statement of what it shows, and the two should agree. I have seen too many slides where the caption claims a trend the chart itself does not obviously support. Order slides from primary to secondary results, not in the order you happened to run the analyses. Statistical detail belongs on the slide or in the appendix, at your judgment, but the headline finding belongs in a sentence, not a p-value.

Discussion: interpretation, not restatement

The discussion is not a second results slide. Its job is to say what the results mean, honestly, including what they do not show. A discussion slide that only restates the results in different words has skipped its actual job. State the interpretation, then state at least one real limitation, not a token, ritual limitation like “small sample size,” but the limitation that actually worries you most about the claim you are making.

Future work: the next real move, not a wish list

One slide, a small number of items, each one a plausible next step someone could actually start soon. “Explore other applications” is not future work, it is a shrug. A concrete next step names what would change, why it follows from what you just showed, and roughly what it would take to try.

Formatting, concretely

A few numbers, since vague advice like “keep it clean” is hard to act on.

• Body text no smaller than 24pt. Titles large enough to read from the back of a mid-sized room.
• Roughly five to seven lines of text per slide, and no more than about six distinct elements (bullets, numbers, icons, whatever) on any one slide. Past six, the eye stops scanning and starts hunting.
• One idea per slide, budgeted at roughly one minute of speaking time. If a slide needs three minutes to get through, it is really two or three slides wearing one title.
• Title each slide with the takeaway, not the category. A title that states a finding gives the audience something. A title that just names a section, like “Results,” gives them nothing. A skimmer flipping through just your titles should be able to reconstruct your argument.
• One font family for body text, one for titles, possibly a different one, but not three. Sans-serif reads better projected. Pick a high-contrast, colorblind-safe palette. There is always someone in the room who cannot distinguish red from green on a scatter plot.
• Citations on the slide itself in short form, author-year is enough. Full references collected on a final slide or in the appendix.
• Figures large enough that axis labels are legible without squinting, which usually means one figure per slide, not four crammed into quadrants.
• Keep a PDF export of the deck on a USB drive or in your email, separate from whatever is opening it live. Projectors fail, laptops don’t connect, video embeds don’t play on the house machine. None of that should be able to end your talk.

Backup slides: prepare the answers you already know are coming

Keep a short appendix after your last content slide: extra results, a fuller methods diagram, robustness checks, for the questions you can already predict from the anticipation exercise in the companion post. Nobody expects you to present these. They exist so that when someone asks, you can say “good question” and actually turn to an answer instead of improvising one from memory.

The test for the whole deck

When the outline is done, run it once, start to finish, and ask whether each slide makes the next slide necessary. If a slide could be deleted and the story would still hold together, it was never load-bearing, and a talk with no dead weight in it is, almost by definition, a talk that respects the audience’s time. That is really the whole goal here: not decoration, not a template, just a clear, honest argument that someone else could follow with their eyes closed.

Where to go deeper

None of this is invented from scratch. It is my synthesis of what I have watched work in the room, filtered through advice I trust. If you want the primary sources instead of my summary of them:

• Princeton’s PCUR guide to research presentations is the clearest short statement of the “narrative arc” idea behind the spine of this post: motivation as setup, results as climax, discussion as resolution.
• “Ten Simple Rules for Effective Presentation Slides” (PLOS Computational Biology) is where several of the concrete numbers above come from: one idea per slide, roughly one minute per slide, informative headlines instead of generic ones, six or fewer elements per slide, colorblind-safe palettes, and a backup PDF in case the room’s technology fails you.
• UCSB’s Grad Slam presentation resources page is aimed at short-format talks, three-minute-thesis style, but its emphasis on ruthless simplicity and rehearsing in front of a real audience, not just a mirror, applies just as well to a fifteen-minute talk.
• For book-length treatments: Scientific Presentation Skills by Martins Zaumanis builds a five-part framework around getting a talk cited and remembered, with real conference case studies. Academic Writing, Visualization, Presentation, and Publishing of Research (Routledge) treats the slide deck as one stage in a pipeline that starts at your dissertation chapters and ends at peer review, a useful frame if you are in the social sciences. Academic Conference Presentations: A Step-By-Step Guide by Mark R. Freiermuth is written specifically for non-native English speakers navigating abstract submission, group presentations, and Q&A at international conferences.
• On video, two are worth your twenty minutes: a STEM-focused talk on structuring a presentation around six core slides (title, background, condensed methods, results, discussion, impact), and a talk on opening a PhD presentation the way a TED talk opens, with a concrete scene before the literature, not after it.
• If what you are building is a poster, not a slide deck, everything above still applies in spirit: one argument, one takeaway per section, ruthless simplicity. But the medium is different enough that it deserves its own reading. UNC’s graduate school poster tips page is a good entry point. It curates the “Better Posters” movement’s critique of the traditional dense academic poster, plus concrete guidance on colorblind-safe design and font sizing that transfers directly from the slide advice above.

I have not vetted every claim in the books above line by line. Treat them as a reading list, not a citation, and judge them the same critical way I am asking you to judge everything else that goes on your slides.