One of the things that I truly dig about Ryan is his utter willingness to try anything. Goofy? Serious? Whatever? It’s all in play for him. What keeps people on pages longer he’ll do.
I’m now free to do that. I was constrained by subordinating everything to someone else’s “best practices.” The Caples stuff, other people’s methods, best practices, and the Fortin stuff that you see working. I am Japan. Like Bawld Guy. Take ideas, appropriate…make ’em mine, lather rinse repeat, and knock out another 5 of my 10,000 hours.
No more need simply copy. Enter A/B Testing. You can create a loop that corrects itself. By not having your ego involved, by subordinating EVERYTHING to effectiveness, you can try ANYTHING and see what’s what. Wanna see if pink hippos sell? Go.
Blog Consultant Michael Martine pointed this killer video out, and for those of us using WordPress and a Theme of some type that allows page level layout changes (for color scheme and suchlike) this is the cool.
Now, I can do 2 things: see if pink hippos sell , and see what sells better than what else.
Questions I am going to address:
- Does a highly produced video sell better or worse than a “Garage bandy” deal?
- Does 16×9 kill 4×3 like I think it does?
- Does asking for a sale work better than asking for an opt in?
- Does asking for a sale AND an opt in lower the chances of either one happening?
- Does leading or closing with testimonials work better?
- Do testimonials work at all?
- Should I have dense or sparse sidebars for the purpose of getting opts?
Heady stuff, and stuff that can let us run experiments to test it, and guard our marketing dollars. This is an utter blast, if testing is part of what you do on a regular basis, you can constantly improve your marketing.
When you learn what people respond to in marketing…
…you can improve your salesmanship.
That’s why we do what we do.
How to do A/B Testing with WordPress from Carsonified on Vimeo.
Todd Carpenter says:
How long would you run an A/B test before deciding which version is better? I’m not talking time as much as page views. What size sample equates to a result you can trust?
October 26, 2009 — 5:28 pm
Genuine Chris Johnson says:
Yeah, if there’s no clear winner, I’m not slick enough to deterimine type I or II errors. A clear winner:
http://flatratewebjobs.com/getablog gets 22% opt ins from PPC traffic. Without the sound effects and cheesiness, it gets less than 3%.
http://flatrate.s3.amazonaws.com/4bp.flv <---losing video, by an unfunny margin. I liked the losing video better. That was after 200 hits. Now I'm going to test out a clean page with no "blog" elements using Thesis's "move shit and pages around" engine.
October 26, 2009 — 5:43 pm