On a web site I am working on right now, I am at the stage of just creating many pages targeting different terms and placing different sales styles to them. Then, of course, to watch the logs like a hawk to see which pages bring the visitors to the site and which ones convert.
My plan was to streamline the site in the end and just leave behind what works good and remove the rest of the 'slacker' pages. I usually advise people to leave pages up once they are created, but thought since this was my own project I might streamline the site for ease of maintenance. I think my plans may have changed a little now.
I spend quite a bit of time today just reading different sales and marketing pages on the net's endless supply. I stumbled across some very intersting goodies. Even read an article at SitePoint with Brett's 'theme pyramid' quoted.
This was particularly interesting to me though. I don't know how old this article [ideamarketers.com] is but it sure woke me up to the truth.
I think I'll stick to my own advice from now on...
In direct mail it is the rule to test several versions in small groups before doing a mass mailing. Before I came to the web, I worked for a company that used direct mail, and this kind of comparison testing was a constant.
I've tried to use the same principle on the web, but it's not always easy to ensure the samples are comparable. This is especially true when search engine returns are the only source of new traffic. Sometimes a site is willing to alternate the clickthrough URL on a banner ad, and that's pretty close to ideal for testing purposes, but of course it costs $$$$ over using unpaid search returns. Overture is fertile ground for comparisons, because you can change the clickthrough URL on the very same ad. The only variation introduced into your sample is the time each ad was visible.
What I do is create at least two versions of a sales pitch page and then compare the results. I try to only vary one significant thing between the two versions -- the graphic, or the headline, or a subhead for instance. The one with the best results stays, and the other gets changed.
This way, the sales pitch gets fine-tuned over time to a highly effective level. Because the changes are being dictated by actual numbers rather than subjective opinion, continual improvement is almost guaranteed.