Release a game in the past year or so.  I wanted to focus on a recently completed project to ensure my talk was relevant.  I almost submitted a talk after the Halo 5: Guardians beta was released, but I decided waiting another year for the full release was best.

Ensure the game gets critical acclaim and/or good press coverage.  I figured name recognition wasn't going to hurt.  Halo checked that box.

Ensure I have the time it takes to commit to this.  I'll break this down per phase, but I knew this was gonna be a big effort.

Ensure I've had enough practice presenting in front of large groups.  I tried making up for this between submitting and presenting, but this is one area I probably could have been better prepared for.

Ensure I know/trust people that can commit to proofreading and giving feedback.  There was no way I could do this alone.  It's a good thing I have awesome old colleagues that have the patience to help me out when I need it.

Play a significant role in a topic I'd like to talk about.  I needed to know the material inside and out, but maybe even more difficult was feeling comfortable taking significant credit for it.  I was really self conscious about that going into my first submission.

Choose a topic that has mass appeal.  I wanted to ensure that pretty much any game developer, no matter what their discipline, would benefit from and enjoy my talk.  For better or worse, I think game design topics get away with that, inherently.

Focus on the details.  There are GDC talks that I haven't enjoyed.  Usually, it's because the talk just scratched the surface on a topic I wanted to dive deeper into.  Jamie Griesemer's "Design In Detail" talks solidified that for me.  They are my favorite.  Watch this one, and then all the rest of them.

Present actionable takeaways.  This was really important for me.  My favorite talks ALWAYS have clear, actionable advice.  Like, I could go back to my desk and immediately start implementing new practices to improve my work.

Less is more.  After getting a first pass of the full presentation completed, I never added additional material.  I only removed material in favor of ensuring I had time to get into enough detail on the areas that were most important to me.

Using Google Slides for this stage was really nice.  Eventually you'll need to create a proper PowerPoint presentation using the template provided, but at this stage using Google Slides is great cause you can work on the talk from anywhere.  I hated having to keep track of my latest file between all the computers I use after switching to PowerPoint.

Ask for feedback early.  It can be really nerve racking opening yourself up to criticism on something like this, but just bite the  bullet and do it sooner rather than later.  You don't want to go down the wrong path for too long and then be told it's not working out.

Remove as much text from the slides as you can.  I got this feedback a few times.  I stripped out every word I could by the end.  It's less overwhelming for the audience.  A wall of text can be intimidating/confusing.

Present one bullet point at a time.  Don't display the full list all at once.  Spoon feeding one at a time is a little more manageable on the receiving end.

Keep the slides engaging (i.e. don't let them be static for too long).  Maybe it's A.D.D., maybe it's the internet or maybe it's just people, but we get distracted.  One way I tried to combat this was to have a constant, slow drip feed of animations in the presentation to keep people focused on what was going to pop up next.

Use animated gifs, but don't keep them up for too long.  If a picture paints a thousand words, animated gifs can paint a million.  But, don't leave them up for too long, otherwise that's all people will be able to look at and they'll stop listening to you.

Record yourself giving the talk...then do it again...and again.