Many speakers have big problems with timing, especially in a conference setting when you cannot go long. Though I have practiced talks against a clock, this does not really help me when I'm in the session - if a question takes more time than I should have allowed or I ramble, then the time I took in practice means nothing. I'm sure other speakers have ways to deal with this, but I inadvertently came up with something on my own that helped me enormously last week and wanted to share it. It is probably not a new idea, but it worked well for me since I thought of it rather than trying to follow someone else's suggestion.

I had ended up with one of the one hour session slots at the end of DevConnections for my WSE3.0 Overview talk and knew that posed a problem.

I looked at the powerpoint deck and divided the presentation up by topic. Then, off the top of my head, wrote down how many minutes I thought each topic (including demos) should take. Luckily, this added up to 55 minutes!

Then in a notebook (notebook is a tip I got from Ingo Rammer) I wrote down a name for each section and then, based on how long I thought the previous section should take, what time it should be when I started that section.

It looked like this:

The session was supposed to end at 3:45. I knew I was cutting it very close for Q&A, but since it was a short session, I told them at the beginning that we would not have a lot of time for Q&A and could continue it in the hallway or online afterward.

So this worked for me like a charm. I had my little travel clock right on top of the notebook and it was easy enough for me to remember to take a very quick look over there as I started each section to see how I was doing. In this way I was able to determine if I needed to speed up or if I was okay.

I wish I had come up with this prior to TechEd South Africa where we had one hour slots but were told to leave 15 minutes for Q&A, making the presentations only 45 minutes long. But now I know I can do this from now on and hopefully it will help someone else.