Doing Meetup and conference talks has been one of the most important boosters for my career. I’ve also meet many amazing people and visited some amazing places, as extra benefits. I strongly recommend that everyone give it a try.
I also have the privilege of being on the program committees for several conferences. Hence, I review a lot of talk proposals. Most are badly written, while many are excellent. Thanks to some prodding (thanks, Vitaly!), here are my suggestions for improving your proposals and thereby improving your chances for acceptance. I’ll also throw in some tips for making the talk a success, once you get past jerks like me.
You have Two, Maybe Three Sentences to Grab My Attention
The first is the title and the second and third are the opening for your description. Too many proposals start with a bad title and a long paragraph regurgitating the same boilerplate that everybody already knows.
Think back to the last time you went to a conference. How did you pick the next talk to attend? You took 10 seconds to scan the list of titles and you read one or two opening sentences of the descriptions, if the title looked interesting. Now look at the proposal you’re writing. Would you just skim past it?
Keep in mind that you are not a professional comedian. You are not my drinking buddy. Keep it professional and concise. Convince me that you’re the indispensable expert with whom I want to spend two minutes now and maybe 30–60 minutes at the conference.
A made-up example:
Title: Running Python-based Machine Learning Models in a JVM Production Environment
Like most organizations, E Corp data scientists prefer Python-based tools, like scikit-learn, while our production systems are JVM-based, like Spark. After many experiments and dead ends, we now use three techniques to run Python-based machine learning applications.
I’ll continue below. (I’m not a Python ML expert, so some details in this example are made up and likely wrong or will be when I continue below. Caveat Emptor.)
Who Is Your Audience?
Imagine we’re standing in an elevator and you’re proposing your talk. Imagine we’re in a room and you’re giving your talk. What do you expect me to already know or not know? What concrete ideas do you want them to take away from your talk? In general, what do you know about the audience that will be at this conference? Which subset are you trying to reach?
Some conference proposal forms require you to specify the intended audience and outcomes for attendees. I wish all did. Use this reminder to crystalize your proposal.
How Much Context Will I Need to Read First?
Continuing with the boilerplate and audience topics, if you’re proposing a talk on a currently-hot topic, chances are everyone already knows the issues that need solving, so don’t state the obvious. Instead, state the particular problem your talk addresses. Then state your solution. In the example above, I tried to state the specific problem as quickly as possible, then got to the main point of the proposal.
Focus on What’s Unique about Your Talk
Note that I used the singular in the previous paragraph; “particular problem”. What is the one topic that you and only you can explain to the world?
Too many proposals pile on everything they can think of. How much time will you have for your talk? Of all that stuff you mentioned, how much is really central to your topic? What can your audience learn from someone else? Could some of your great content go into a supplemental blog post instead? I’ve added “bonus slides” to the talk itself with this kind of extra content.
Unique and valuable doesn’t necessarily mean novel. Your talk might recap established ideas for beginners or compare and contrast several well-known approaches. These talks require a special set of skills. They can be difficult, because it’s too easy to just show a boring list of this vs. that. From your unique experience, what should I know, so I skip the pain you went through to reach your level of knowledge? Since your choices might not be right for my situation, what techniques can you teach me so that I can quickly do my own evaluations? What acceptance criteria do you recommend so I reach the best concrete outcome?
A related problem, it’s disrespectful of the reviewers time to submit several, almost-identical proposals on the same topic, hoping that one variant sticks. This backfires; I’m more inclined to downgrade all of them. Please submit one, well-written proposal per topic.
Keep It Concrete and Concise
You also convey uniqueness and naturally attract interest by being very specific. If you say “Come learn how MonstersOfAdTech serves ads”, I’m already reading the next proposal.
You know you shouldn’t put bullet points in your slides, but they can be useful in your proposal. I can scan them quickly, if they are concise.
Continuing with my previous example:
Specifically, we do the following:
* Run scikit-learn code using PySpark
* Use PyOnJVMClusters to distribute and coordinate execution of scikit-learn processes across multiple nodes
* Use PyOnGPUs to translate Python and run code using CUDA
I’ll discuss why we ended up with three approaches, instead of just one, and the circumstances where each technique is appropriate, such as application types and performance considerations. I also discuss the challenges and unsolved issues that require more work.
About Live Demos
Live demos can be cool. They can grab attention. They can also fail. They can also be surprisingly boring.
The most valuable use of a live demo is to show some behavior evolving in time. Note how the response evolves in response to the changing inputs. Note that the spike in input is effectively smoothed out by the application, so downstream consumers see more uniform traffic flow.
Is Clicking a GUI or typing commands in a terminal really that exciting and necessary? What can people see from the back of the room that makes them say, “Wow!”
Avoid Product Pitches
A lot of proposals come from people who would like you to buy their product or service. This is understandable. Travel expenses for most speakers are paid out of Marketing budgets. However, an obvious product pitch will rarely be accepted. Instead, sell indirectly. By discussing the technical problems you solved and how you solved them, you both establish your street cred and you still provide useful information for people who won’t or can’t use your product or service.
Similarly, if you’re discussing an in-house framework that isn’t open source, you must give the listener useful advice they can use. If your framework requires an ad hoc scripting language, weird process steps, etc., forget it.
Don’t Let the PR Department Write Your Proposal
“Come hear our self-absorbed CEO tell you why we’re great and you’re not.”
Write your own proposals. You know what you need to say better than the PR flaks.
Don’t Quote Gartner
You are proposing a talk at a conference where people come to learn what’s new and what’s hot. If you’re quoting Gartner or any other research firm, you’re telling me what’s already mainstream.
Also, don’t make grandiose claims. I demand hazard pay every time I have to read a line like, “E Corp is transforming Artificial Intelligence into Real Intelligence.” (Not too far from a line I read this morning in a talk proposal…)
Learn from the Masters
We’ve known since at least the time of Aristotle that a good, compelling story uses tension and release; it sets up a problem, then provides a resolution.
Read the titles and descriptions of the talks that were accepted to previous conferences. Those proposals obviously worked. Use a little caution; some proposals are accepted for one reason or another, despite flaws.
Watch the talk videos and read the slides for those talks that received high review scores from attendees or lots of positive comments on social media.
If you have the time, read a few bad proposals (if you can find them somewhere) and watch a few poor talks, too. What makes you react negatively? What would you do differently?
Use Good English
Obviously, many proposals come from people who are not native English speakers. Given the prevalence of excellent spelling and grammar checkers, I’m surprised how many proposals have glaring spelling and grammar mistakes. If you can ask a native speaker to proofread your proposal, do that, too.
Actually, always ask others to review your proposal! They’ll catch confusing wording, assumptions you made that aren’t obvious to the reader, and so forth.
I hope these suggestions make your next conference proposal a success. I look forward to reading it.