Reading Techmeme on Tuesday, I came across the Compete blog, and these fabulous images related to the upcoming presidential election:

If you’re interested in their analysis, read the full post; it’s excellent stuff. It’s also a great example of the speed with which we move these days. Michele’s earlier post about Twittering with John Edwards is another.

The Compete folks predicted, based on website traffic, that Mitt Romney would win the GOP nomination. I love their spirit, and the visuals are damn cool. But as this is the Conversion Rate Marketing Blog, I have to admit, Conversion would be a far greater predictor than Traffic.

I know we’re biased but, at the end of the day, when you’re running for President, what you care about first and foremost is votes (much like when you’re running your business, dollars and delighted customers come first). People voting is about an audience taking an action based on their own motivations. And that’s exactly how we’ve defined conversion for the past decade.

The election itself measures macro-conversions (i.e., votes) but the predictive model can only measure micro-conversions (e.g., donations, volunteering, etc.). Just like smart marketers plan a persuasive system to predict sales, one could plan a persuasive system to predict votes. It’s simple, but that doesn’t mean it’s easy

1. First, you’d define the audience, using an intelligent framework like Jungian archetypes or Myers-Briggs (define – find your type).
2. Then you’d consider the micro-actions you’d want them to take; those from an earlier-stage decision than voting. For instance, offering an email address as a way of communicating, signing up for a Twitter update, making a financial donation, or volunteering your time.
3. Once you’ve decided on the micro-action conversions, plan an experience to facilitate these actions you’ve laid out for your audience. The key is in understanding the individual’s motivations for taking the action you’d like her to take.

Some of this information is publicly reported, specifically fundraising totals. Fundraising shows a different prediction than the traffic alone; namely, Rudy getting the nod to partake in the 2008 general election. Obviously, campaign donations represent only one scenario, of which there are many.

Anybody know any sources that track the other potential scenarios? I’d love to assemble a predictive model based on the wisdom of crowds.

[Editor's note: For more online campaign analysis, stay tuned for "If Clicks Were Votes" -- President Obama?"]