How to give audience-focused talks

Tips on giving audience-focused presentations

Talk are important for disseminating results, creating professional connections, being remembered, and ultimately career progression. Although content is king, and people will come to your talks to hear about your science and thus the content you have to offer, a talk has many entertainment aspects that scientists are rarely trained on.

Moreover, before our first talk we often mostly see slides from in-classroom teaching (which serve a different purpose and are usually not necessarily good examples for a presentation).

I outline here some thoughts and rules of thumb learned over the years, from my own talks and others.

N.B.: If I get totally lost while someone gives a talk, I try to pay attention to talk design and structure, delivery of the speaker, and all secondary aspects: even if the content is lost to me, I can at least try to learn something on this secondary level.

As any set of rules: it's made to be broken. Nevertheless, having these in mind helped me a lot in designing talks, and whenever I "break" one of my rules, I now do it consciously knowing why.

The various kind of learners

The advise below comes from recognizing that in your audience there will be a mixture of people with different preferred learning modality:

auditory (people that prefer to listen)
visual (people that prefer to see/read)
synaesthetic (people that learn from the combination of stimuli)

You yourself probably fit in one of these categories, but when you give a presentation you want to serve the content to all these categories.

The main take home point

You should put a lot of effort in preparing your talks and always practice multiple times the delivery.

Ask peers for feedback and help. Give peers feedback and help. Learn from each other!

Building process

To state the obvious: you give talks not for yourself but for the audience. During the entire preparation of the talk (from conception, to preparing support material, to the delivery) you should think more about the audience than yourself.

N.B.: This also means that you are allowed all tricks/supporting materials that help you deliver a good presentation for your audience. You can have notes (presenter notes, cue cards, etc).

Whatever you show to the audience is for them, not you. Do not put lots of text on the slides because it helps you remember: the slides are for the audience, if you need to remember something, make a note available only to you (there is no shame, professional speakers like anchors and politicians speak with teleprompters and/or notes).

Preliminary questions

Ask yourself these questions before starting to put together a talk!!

Who is your audience?
What do you want to achieve with this talk to them specifically?
How long is the talk going to be?

For example: what is the seniority level of your audience, what are they interested in or what do they expect (based on what kind of talk it is).

And in terms of what do you want to achieve, there could be many things:

present ongoing research to gather feedback
advertise new results
summarize a research field
educate a non-expert audience
introduce a discussion section (e.g., open ended questions)
get a job offer!
…

Once you have an idea of what the audience expects and what you want to achieve/obtain from this audience, then you can start making your talk aiming for the duration allocated to you.

The content and core messages

Content is the most important thing of your talk: you can give a good stage performance without almost any content, but for a scientific talk, that is unlikely to go well and achieve your goals (see question 2 above).

Your content should be organized around "core messages". What your core message is depends on the questions above (who is your audience? what do they want? what do you want?): it should be extra clear to you what the core message is, and you should work to make sure after each slide this is clear to your audience as well.

You can have at most one per 15-20min talk (and maybe up to 3 or 4 for a one-hour seminar): this is the main take-home point you want to leave the audience with. If they should remember one thing, it should be the core message.

During the talk you need repeat this message multiple times (rule of thumb: at least three times for a ∼15min talk). You can add "variations on the theme": for example adding details and specificity, or announcing that's where you are going at the beginning, building up to the core message, and then repeating it in your conclusions.

Do not try to pack many messages in one talk: for most of your audience this is not going to work. They are not as familiar as you are with your content, if in ∼15min you manage to explain well one thing, that's a great success (how long did it take you to understand that?)

The storyline

Human beings are good at narratives, following a story with a main character. This may be a result of evolution, and you can exploit it to your advantage.

You should trim and organize the content around and in support of the core message(s). Based on what you want to achieve, and who is your audience, take time to select and organize your content in the most coherent, logical and non-surprising way. Think of an "hour-glass structure": start broad (give context), progressively narrow to your core message(s), and then broaden back for the conclusions. Sometimes this will naturally follow the structure of a paper (e.g., when the aim is delivering new results and the core message is the main result), but not always.

Ask yourself if you can frame your talk into a coherent logical narrative based on one or few characters (e.g., the star is born, evolves, dies), or many little sub-stories (like chapters). The storyline should deliver the core message (and allow for multiple repetitions of it), and "secondary" smaller message should be functional to building up or corroborate the main core message.

N.B.: The storyline does not need to be particularly creative per se, it is not the content, a simple and clear storyline is better than something with twists and turns!

Do not try to "hide" the story: instead you can make it apparent! This will help your audience keep track of where we are in the talk, where we came from, and where we are going. This clarity will help the audience focus on the content without the need to think about the storyline. This will reduce the cognitive load and ultimately help them understand what you want to communicate.

Sometimes people even go over-board and make the story too explicit which can be a bit cheesy, but can work (it really depends on your style as a presenter).

A common technique to visualize core message(s) and possible storyline building blocks is making mind maps. These can be useful to connect and organize topics and subtopics around the core message. Personally, it never worked super well for me. Instead, I typically start by the slides with the core messages and or a table of content (which I may remove later) to map how to get to the core message slide (N.B.: this is generally discouraged, since it can lead to procrastinating by fiddling with slides instead of focusing on the big picture, the core message, the logical progression).

Supporting material

It is very common to give talks with slides, so I'll focus primarily on them, but note that many advises apply also to a black/white board talk (same as above!).

There should be one main point per slide, and not more than 1 slide per minute on average (reasonable rule of thumb). Keep it as simple as possible while still supporting the points being made.

The slides should also be ordered following the storyline. Avoid content-less slides (e.g., title: Outline, slide: list of bullet points saying Intro, Method, Results, Conclusions. Nobody needs that slide, not your audience, not you. If instead your outline is non-trivial and you use the slide to make the storyline apparent, that can make sense).

Place-holder slides to emphasize the story-line, give a break to the audience (and allow them to jump back on board if they got lost) are good to use explicitly: those are content-less maybe but have a clear purpose in managing the audience attention!

Avoid complicated crowded slides

The more you fit on the slide, the more content there is for the audience to parse while they also try to listen to you, the more you confuse your audience. Generally, people are not good at reading and listening at the same time.

Slides should not be able to give the talk without you. Slides are just a support for you to deliver the content, they should complement and support, not repeat or substitute what you say. The delivery is from the combination of you + your supporting material, the latter needs not to speak for himself.

Do not use the slides as your reminders, there are other tools for that (presenter notes, cue cards, writing on your hands). Instead, plan to practice enough your talk so that you don't need reminders! I usually rely mostly on the latter: practice, practice, practice!

N.B.: presenter notes are great help, but sometimes a room doesn't have a presenter screen, or you can't use your laptop. They are not a substitute for practicing.

Do not show anything you don't need: remove legends that are distracting, clean figures from lines you don't discuss, simplify and distill every word, image, line on your slides. Make your axes labels and tick labels readable (modify the figures with your presentation tool to re-write these if needed). The audience will already be very busy trying to absorb the new content you deliver, don't make them do the "cleanup" job. Once again, you make the slides for them not you: if they as your audience don't need something on the slide, remove it!

Similarly, unless serving a purpose, avoid animated transitions: they are just a distraction and come out as childish and unprofessional.

Keep the slide style as simple and uniform as possible (I personally go for white or black plain background): the fun, exciting part of your talk shouldn't be how pretty and/or complex your supporting material is, it should be how cool your content is and how well you explain it!
Layer the content

Not everything on the slide has the same level of priority. Certain things are naturally "high priority", meaning the eyes of the audience will go there for sure. For example: the top part of the slide (where the title may typically be). You can exploit that! Important stuff goes there. If there is one sentence or longer string of text, place it there.

You can guide the eye of the readers with font size (bigger → higher priority), contrast (more contrast → high priority), position (center higher priority than margins, top higher priority than bottom, left and right may be culturally dependent and influenced by how you construct the slide).

As an example of secondary information: consider citations. In your scientific presentations, you should give credits to the relevant literature and also give pointers (link, arXiv numbers, QR codes) to your work, so citations belong on the slide. However, only a fraction of the audience will be interested, they shouldn't draw the most attention! I always put citations close to where on the slide I'm using them (e.g., next or in a figure I borrow), in gray text (lower contrast) and tiny (but still legible) font. Do not put all your citations lumped in one slide at the end of the deck, then none knows what is cited for what!

You can use colors, fonts, and shapes to layer content, but be also aware of the limitations of the human brain: a rule of thumb is never more than 6 colors (that's already a lot, aim for 2-3 max), and generally size > colors > shapes in terms of perceptual importance.
Avoid bullet point lists

Everybody uses them (I'm also guilty!) and they are terrible. They are easy to make for the speaker, but typically hard to follow for the audience. Remember, you're making the talk for the audience, not for yourself! Most of the time, whenever I catch myself making bullet point lists, I try to spend time to see if I can re-organize the slide to write each bullet next to something it refers to.

Bad example:

Better example:

A slide with an image in half the slide and a list of bullets describing particular elements of the image on the other half can be reorganized much better making the image full size and write text (maybe with a semitransparent background) next to the portions of the image being described. It's more work for you, but will help the audience in a subtle way they may not even notice – but the collection of these subtleties can make a talk stand out!

Related: avoid teasing by revealing one bullet at a time. This is a way for the speaker to force the audience to listen to the current point. The audience will have a range of paces of comprehension, people learn going back and forth and weaving their understanding with your words and content, do not force them into passive listening by taking away their ability to wander with their mind.
Less is more: avoid too much text

Typically only the title of the slide will be fully read by the audience (and I advise making it "journal" style remove articles and prepositions).

If you need more text, the rest should be individual words, short expressions. A picture is worth 1000 words, then why would you put words on a visual medium like a slide?! The fewer words the better (zero = great!).

Movies can be great (especially to show things evolving in time!), but it can be tricky to balance the audience attention between your voice and the moving images (typically: your voice will lose). Another potential issue is if you cannot use your laptop.

The text you are tempted to write, in most cases, is what you should say on the slide, then it shouldn't be written for you to remember on the slide, instead you should practice it. Remember: your slides are not meant to speak for themselves!

N.B.: you can look at presentations of products from famous CEOs whose job is to get the stock market hyped about the product and drive the value of the stocks higher. You'll find very few words on their slides!

Similarly, try to not show equations (symbolic representations of math are not very good at being grasped on-the-fly by the audience). You want your audience to understand something, not to just show off how smart you are and how much you can complicate things, right? If you really need an equation, ask yourself if you can simplify it (maybe drop the constants and make it a proportionality, maybe write just \(\mathcal{S}\) for the complicated source term that you're not getting into, etc.). For most people, a plot can illustrate quantitative relationships better than an equation at glance.

Delivery

This is the aspect to practice, practice, practice. By practice I mean put your slides in presentation mode, time yourself, pretend there is an audience you are talking to (speak as loud as you should during the actual talk, if you stumble, get it back together and don't stop).

If practicing you are too long, don't speak faster, cut something! Also keep in mind in the first practice round you'll do many more "uhm, ehm, ohm…" while you come up with the phrasing, so it's often the longest).

I always still do this multiple times per talk almost regardless of how many times I've done it before. You'll know you've practiced enough when you can give the full talk without ever looking at the slides and almost not thinking about what you're saying (ok, I don't always reach this level, but until the end of my PhD I did reach it!).

During the delivery there are many aspect that come to play, maybe subconsciously:

body language (you don't want to be defensive nor all over the place, you don't want it to become a distraction but also you want to come across as natural and engaging)
position (w.r.t. the slide/board, the audience – including online audience)
voice tone (you should project! Practice without a microphone, ask at the beginning if everyone can hear you, pronounce as clearly as you can)

While you give a talk, you shouldn't be thinking too much about any of these though: you should have practiced enough that most of these things are engraved in your subconscious and come out "naturally" without you thinking about them.

The right things to do are not necessarily what comes natural to you: it's ok to make up a "talk-delivering persona" for whom those things have to become natural. This persona should be inviting, engaging, captivating for the audience and capable of delivering complicated messages in simple forms adapted to the audience itself.

Among those (not in any particular order):

Posture: your posture should not comunicate a defensive approach (e.g., crossed arms): observe news presenter (the ones standing), they all have a similar posture with their hands holding each other in front of them roughly at the height of their belly, standing comfortably (legs spread as much as shoulders) sometimes putting weights more on the tip of the feet.
Speak clearly. Project your voice, talk to the audience (not with your back to them), look around the room (try to make it natural, not a robotic scan), cross eyes with the audience (or, if that's to scary, stare in between two people pretending to engage with someone behind them). Imagine that each sentence is something you deliver to one person in the audience (but loud and clear, you're actually delivering it to everybody): this sentence goes there to a specific person, and that sentence goes up there in the back, and so on. This will emphasize that you're speaking with this audience and not at whomever is passing by accident. You can practice this too: in your room where you practice you should imagine the audience and start doing this already (e.g., this sentence I look at the lamp, this other one to the corner) and practice so it's not an unnatural robotic scan of the room.
Welcome the audience: this establishes a connection between you (the speaker) and them (the audience) and signals that you are there for them. Maybe thank them to have come to listen to you. A clear example of welcoming the audience to establish a connection is practically any talk show host:

Instead, I advise against "thanking the organizers for giving you a chance …": its formulaic, boring, and self-referential. Thank the organizers by contributing to the success of what they organized and deliver a great talk instead!
Pitch the talk to the audience, by this I explicitly mean not to the most expert person in the room. If you don't know who is the audience (you should have asked the organizers at the beginning of your preparation!), a safe bet is pitching to first year masters/PhD students: assume some general knowledge but no specific or technical knowledge.
Make your gestures supporting what you are saying. You can move around, but pacing to release your stress becomes a distraction: if you move try to make it related to the point you are making (e.g., if you compare and contrast two things, speak of one on one side and of the other on the other side, your whole body can act as the "placeholder"). Same with your hands gesture or anything else: make them meaningful, not distracting. Synaesthetic learners will particularly appreciate this.
Never ever go over time. At time + 30 sec nobody will be listening to you anymore. Any content delivered is lost. You want to get to the end of what you prepared, but that's for your satisfaction, not for the audience. Instead, if you see you're going long, cut something. N.B.: you need to prepare for this, maybe the last chapter of the storyline can be skipped? Skipping ahead is not needed if you timed yourself well, but if needed, it looks more professional and mature to do that than rushing through your slides without anyone understanding. That remaining minute is best used for Q&A.
Wrap up with a conclusion or summary. Which is more appropriate depends on the context, the audience, and what you want. Maybe a summary and ending with your question to the audience is what you want (e.g., to start a discussion section at a conference), maybe a summary highlighting your recent results (or how they relate to each other) is what you need for self-advertisement purpose.
In the Q&A part, repeat the question to the person asking before giving your answer. This forces you to engage as a human with the audience, rather than being a "talk delivery machine", it brings the entire audience on the same page (especially the people on zoom who didn't hear anything because the person didn't wait for the mic), and buys you time to answer while making sure you understand the question. Only positive things. Receiving questions means the audience was engaged and listened: it's a sign of success! You shouldn't be scared of questions, most of the time they are genuine curiosities, and you know much more than your audience about your work. And it's ok not to know. The answer "I don't know" (or even better, "I don't know, on the spot I could speculate hypothesis") is totally ok and much better than saying confidently something completely wrong (you're not an LLM!)