A conversation on Twitter got me thinking. That's kind of rare. Thinking, I mean. At least for me.

In my time, I've given a lot of demos. I am not a demo god. Don Box is a demo god. Scott Guthrie simply exists on a separate plane. I mostly note what they do from the sidelines. I've picked up a few things of note. In the words of Michael Gartenberg, "Like most worthwhile things, good demos are hard to do, look easy from afar, require lots of practice to perfect."

So what have I observed the best demoers do (either explicitly or implicitly)? Seven things:

First, know your audience. Noobs? CEOs? Noob CEOs? Devs with more experience in the product than you'll ever have? Friendly? Hostile? Bored? Are you giving the post-lunch demo? Knowing the audience means that you'll have a higher likelihood of connecting with them. Or at least not insulting them.

Second, know your product. This is especially critical if a) there is the chance that an audience member will ask you a question or b) the product is in one of those awkward beta states where it may stick four paws in the air on you at any given moment. You can skimp here if you know this is a one-off, but it's not recommended.

Third, know what you're trying to get across to your audience. Is this demo an:

• Introduction to a concept. Sometimes the hardest demos are the ones where you're introducing the audience to a new (and potentially abstract) concept. Imagine being the first person to ever demo object-relational mapping.
• Introduction to a multi-product scenario. Second on the difficulty list: a demo involving multiple products in which you're introducing people to at least one of them for the first time.
• Introduction to the product. These are kind of easy -- the audience knows you're demoing a phone, for example. Your job is just to make it look good. As Jonathan Yarmis pointed out: good products don't need great demoers.
• Introduction to a feature. By the time you're introducing a feature, your audience knows the product. You can generally get pretty in-depth. These can be a lot of fun also.
• Explanation of how to do something specific. How-to demos are a special case: demo as training tool. Rather than dazzling with BS, you're trying to train and inspire. Don Box rocks at these.

Fourth, know your themes. A demo, like any presentation, can push 1, 3, 5, or 7 themes. Don't ask why the odd numbers -- it's a marketing thing. Nobody in marketing ever makes 2 or 4 themes. I'd mock it further, but it works. Pick your themes carefully. The worst demos are the ones that try to make too many themes . Even a knowledgeable viewer will get lost. I've seen demos of my own products that have left me confused. Are you trying to hammer home that this product is incredibly simple to use, or are you demonstrating that the previously impossible is now merely hard? Are you trying to subtly point out that a competitor doesn't do what you do? Are you trying to convey cool and new or secure and enterprise-ready. Knowing your themes dictates the next step.

Fifth, know yourself. Are you serious? Funny? Boring? Quick-talking? Earnest? Cynical? Whatever you are, it'll probably come out in front of the audience. Plan for it, and accept it.

Sixth, know your story. Once you know all that, spend a minute and mentally script the story, taking into consideration especially how point #5 will affect it. Actually, for big demos (like the big multi-product keynote demos) it's more than a minute. Often there's a huge amount of custom development and tweaking to get the demo right. Getting the story -- the actual words-on-the-left-demo-actions-on-the-right script -- written can be critical. But remember: be true to yourself. If you're by nature cynical and caustic, invite the audience into that world (preferably with humor) rather than trying to be an earnest do-gooder.

Sevent, practice. Especially if anything in the chain is new to you, at least one run through to ensure that the system is set up right and that you know where it's likely to fail is key. Many is the time I've not done a run through only to have that awful, "Boy, it's never done that before" moment.

And oddly, there's no eighth point. Must be all that marketing training. Or the fact that my wrists hurt now.