How I gave 5 talks in 10 days - my top 3 tactics for public speaking preparation
In the month of November 2019, I had multiple public speaking engagements within a very short time span - at the tightest, I delivered 5 talks in 10 days. That’s one talk in 2 days! Here are the 3 main tactics I used to be prepared for the different content as well as the presentations themselves.
- The schedule
- Tactic 1 - Do dry runs
- Tactic 2 - Get to an MVP slide deck as soon as possible
- Tactic 3 - Slides do matter, for maximum effect
- How I would improve my talks in the future
The timeline was extremely tight - I had to start creating my content and slides way in advance. Also, on each of the days I had a presentation, I basically didn’t have time to prepare for other presentations as I would get home exhausted. So between Nov 7 and 18, I barely had a moment to drastically change my content, or even to take a break.
- Nov 7 - Modelling Seminar, Bell
- Nov 9 - Dames Making Games
- Nov 14 - PyLadies Toronto
- Nov 16 - PyCon Canada
- Nov 18 - A.I. Socratic Circles - REINFORCE Recommender System
- Nov 28 - A.I. Conference, Bell
Tactic 1 - Do dry runs
Doing many dry runs is the single reason I could finish preparation for the talks on time. This was new to my public speaking preparation routine, because I usually could deliver decent talks by “winging it” or getting prompted by my slide content.
Winging it often worked well enough, because I have pretty good experience speaking from giving tutorials at Rotman business school (as a teaching assistant), as well as multiple technical conference talks in the past on data science or game development. I’ve also been on stage many times since childhood, from speaking competitions to musical performances in piano and guitar. So “winging it” was definitely a skill I slowly developed throughout my experiences, and not made in a day.
However, in the past, I always had enough time to exclusively practice one presentation at a time, so I needed to change up my strategy to ace talks on multiple topics within a short time. This time, getting to the point of mastery where I didn’t have to think too much on stage, was crucial because of multiple topics I had to prepare for.
It’s tempting to not do serious dry runs - if I can deliver the talk by having read the slides thoroughly and thinking about it in my head, why go the extra mile?
The reason is that what makes a talk good is when the speaker has reached the point of mastery. I liken it to if I was preparing for a violin performance, and for weeks, I only stare at the violin and think about the notes I will play. Then, on the day of the performance, I finally pick up the violin for the first time and play the notes out loud. This is basically the same as going into a public speaking engagement without having done dry runs. A skilled violinist could probably sight read a song (like reading content from slides) without making a “mistake”, but it won’t be as good a performance if they haven’t already practiced on the actual violin prior to the performance.
I’m going to break down my dry run schedule for one of these talks, which was about reinforcement learning applied to the YouTube recommender system - a very technical topic. It was presented at A.I. Socratic Circles, which in addition has sections of the talk dedicated to audience questions and discussions. This meant that I had to know the content inside and out, and be able to deal with improvised content (not on the slide deck) that comes up organically during the presentation.
- Nov 5 - dry run with S.M., O.N.
- Nov 12 - dry run with ~5 coworkers
- Nov 13 - dry run with O.N., J.M.
- Nov 14 - dry run with N.P.
- Nov 17 - dry run with A.D.
- Nov 18 - day of the talk!
For this single talk, I did 5 dry runs to 10 people in total. To closely mimic the presentation format, I let them know they could ask questions at any point during my talk. I got a lot of practice answering questions from different people with varying understanding of reinforcement learning and recommender systems, which I had to adjust my explaination depth for on the fly. I’m very grateful to the individuals who lent me their time - each dry run was ~1 hour long!
Without dry runs, I couldn’t have been able to react so well to the questions during the talk. Beyond this case of a highly interactive presentation format, dry runs also helped immensely for talks where I didn’t have to have back and forth with the audience.
For example, at PyCon Canada (largest Python programming language conference in Canada), I spoke to a very technical audience as well. There was a very tight time limit to ensure the conference ran on schedule. So I practiced mutiple dry runs alone with a stop watch and video recording - I could do it twice an evening and note down what parts of my talk to cut or speed up.
If it’s difficult to get so many people to listen to dry runs like my A.I. Socratic Circles talk, I found it useful to record myself on video. I didn’t end up watching the recordings fully at all (and because of my tight schedule I literally didn’t have enough time in the evenings to do so), but having the camera on forced me to give a real dry run (no random pauses to edit slides, or breaks, in order to mimic the true situation).
One thing to clarify - my presentation style is more conversational. I don’t memorize anything and treat presentations like a dialogue where I am explaining the content to a friend, using the slides as a framework. Due to my presentation style, every single time I do a dry run or real presentation, the exact wording I use is different, and I might even add things on the fly. Because of that, dry runs even help me discover ways of explaining a concept clearer!
Tactic 2 - Get to an MVP slide deck as soon as possible
Apart from getting on stage and speaking, I spent so much time on making the content of the slides. Because some of my dry runs were scheduled up to 2 weeks before the respective talk dates, I had to ensure I had some bare bones content to present in advance.
I literally did not have time to make the most polished versions of the slides during those first dry runs, but that’s the beauty of the “lean” methodology of product improvement - getting feedback from an MVP (minimum viable product, a concept popularized by The Lean Startup by Eric Ries*).
Because I used skeletons of slide content, I was able to iterate quickly on feedback before polishing (spending hours creating well-formatted math equations, dragging images around to beautify the layout…). If I had gotten the slides to 100% polish all before the dry runs, and then deleted or added slides according to the feedback, I would have have wasted all the time I took to polish the slides.
Content improvement iteration first, then polish!
Tactic 3 - Slides do matter, for maximum effect
Look - I’m a technical person by training. I used LaTeX slide decks all through grad school so that math equations look pretty. They do not look pretty for presentations outside of the lecture hall, without careful deliberation. This was a blessing and a curse - people in technical circles do not expect slides to be clear, and do not expect speakers to be charismatic. By at least having a bit of the latter… let’s say I was able to get the spotlight more easily. Improving public speaking skills is an investment in yourself worth making!
The whole point of giving any talk is so that the audience can take away an idea or concept that I am explaining and sharing. If either of below situations happen, I might as well have wasted all my time preparing and giving the talk, as well as the audience’s time.
How to waste everyone’s time (do NOTs of public speaking)
Yes, the title of this sub-section is worded strongly, and I do have strong feelings about these points. You’ve encountered speakers like this, I’ve encountered speakers like this, we all have. So the following points should be a universal feeling.
- Have slides so dense with words that the audience struggles to choose between reading the slides or listening to the speaker, to the point they fail to do either, or space out. I’ve heard talks like this where I felt it was a massive waste of everyone’s time. I might as well read the slides at home. This style is common in lecture halls (and why I skipped class so often), but they don’t belong in more public settings or to a mixed (e.g. technical and business mix) setting.
- The speaker reads the slides verbatim - not feeling like a talk. Same as the previous point, I might as well read the slides at home if the talking portion isn’t giving the audience any value for the time they spend physically getting to, and being in that auditorium.
So, what do these points have to do with my tactic of making the slides? My point is that, in order to not waste my own time and the audience’s time, the communication has to be well executed. This means my words and slides should compliment and enhance each other so that the above 2 time-wasters don’t happen. The more the above 2 time-wasters happen, the less the audience can understand my talk, and the less they walk away remembering my key points, which are the reasons I spent all that time preparing for the talks in the first place!
I went to a PrezExpert training, a full day workshop, courtesy of my workplace. For a person that never liked making slides before, it was life-changing. I started to think of slides as complimentary to my talks, not something to get over with as soon as possible. The time spent on slides has incredible return on investment. They really matter for helping the audience understand me, be it a business or technical or mixed topic difficulty.
My main tips for making presentation slides that enhance a talk are:
Use headers to drive home the point of each slide
Instead of using headers as the title of a list of bullet points (e.g. “Benefits of Alternating Least Squares algorithm”, then bullet point - benefit 1, bullet point - benefit 2), basically have your conclusion of the slide in the header or kicker (e.g. “Alternating Least Squares achieves high accuracy with lightweight data”). The header of the slide is precious real estate - if the audience only reads one thing on the slide (happens more often than not), it’s probably the header. So, with well worded headers, even if the audience is spacing out and only reads the header, they can still get the main takeaway!
Treat your slides like a user interface - make it intuitive for the audience
A talk is like a performance; instead of, say, an instrument, you’re doing the performance with words and visuals. The goal of this performance is for the audience to understand some topic, or even to be persuaded. The performance can even be seen as a product in this sense, which is why I treat the slides and talk like an MVP, and ask for feedback aggressively. Following this logic, when creating the product, consisting of slides and words, thinking from the perspective of the user is important! I guess it’s like user interface design and optimization… How can the user can most intuitively form a connection with the product through the interface?
I also read the book, Talk like TED, by Carmine Gallo* to improve how my visual content and words could compliment each other to reach the audience. The book breaks down select TED talks, and so I watched the videos to get isolated observations that I could then use to improve my own talks.
How I would improve my talks in the future
What I did well this time
I feel that during this month of talks, I accomplished a lot, and got a lot of praise for my deliveries. For one technical topic I even got amazing feedback from non-technical folks that they were able to follow the talk! I think that crafting the headers of the slides to have the high level conclusion really helped - even if the business-side folks space out because of the math equations, they can at least remember the reasons for using the algorithm I’m explaining.
In addition, my time management and discipline tactics worked well - I had the image to the left as my phone wallpaper for the entire month. It’s a checklist of milestones such as when the dry runs were scheduled, when I planned to have the MVP slides done, and so on. This really helped me stay on track despite my time being literally squeezed - even a tiny lapse could cause me to fail to prepare thoroughly for a talk.
What I will change next time
Throughout the month, despite never skipping meals or sacrificing sleep (like in undergrad), and keeping a generally normal schedule, my body really took a toll. I fell ill the day before my last talk, throwing up and sleeping horribly as a result. Hence, before that talk, I physically had to take a nap in the conference green room. It was thanks to my dry run preparations that my content delivery didn’t suffer despite my condition. Then, I was sick and bedridden for a week after all the talks had ended. This post-conference or post-exam sickness likely resulted from the sudden relaxation, which affects the immune system - known as the let-down effect.
I mention this as a warning - I would never knowingly sacrifice my health, but like a first time drinker, I learned my limit and will actively avoid pushing it in the future (like most first time drinkers, right?). For example, I can reduce the let-down effect by regularly unwinding instead of letting stress accumulate for so long.
I aggressively collected feedback throughout the month, and had gotten an observation that my energy level tends to start strong, but decreases throughout a talk. Looking back at my last talk of the month, I know I didn’t maintain my high energy until the end, either. It could have been due to my illness that day, but this experience made me more mindful of energy level consistency, so I can improve in the future.
All in all, I wouldn’t trade this month for anything else - I got to speak at some amazing venues and meet so many amazing people. I’m excited to build upon what I learned and continue to deliver better and better talks!