In my book Web Metrics, I advocate the “Try It, Measure It, Tweak It” method of Web site design. While “TIMITI” will never catch on as an acronym, the idea is simple and can have a profound impact on the success of your site.
The premise may be simple, but the implementation is challenging. Still, there's some new technology making the whole thing a lot more intriguing.
Let's say you buy a pay-per-click keyword on a search engine. You direct the people who click to a specific landing page. That page has an astonishingly large number of variables, including these:
- Window size
- Load time
- Background color
- Font styles
- Font sizes
- Over all layout
- Subsection layout
- Tone of copy
- Length of copy
And the list goes on (yes, I'm getting bored, too).
The TIMITI method suggests coming up with a combination of the above that you feel is the best possible. So try it. Put it online and measure the bejeepers out of it.
Measure clickthroughs, pageviews, revenues, and any other reviews you choose to pursue. Then tweak it. Pick one (and only one) variable and change it (and only it).
Then put your new landing page online and try it again... and measure the bejeepers out of it again.
The hard part is keeping your hands, and the hands of your creative agency, the Java programmers in the back room, the receptionist, the CEO and every other Tom, Dick and Harriet off all the other variables. Change only the background color, or just the font and nothing else. That way, over time, you'll find out which variable makes the biggest difference.
However, until modern science gets on the ball with genetic research and cloning, none of us will live long enough to make this method work to best advantage. What to do? Make the machine do the heavy lifting.
Optimost has created a system to do just that, and it does a decent job of explaining the process in the fewest possible terms: “Optimost will develop a testing matrix based on existing creative materials. The Optimost system will then generate up to 1 million permutations of the creative execution. Through sophisticated experimental design techniques, Optimost will then test certain combinations of variables to arrive at the optimal values along each parameter.”
In other words, it's their job to figure out which variables matter, which variable values are improvements and which variables barely move the needle. The more people touching down on your landing page, the faster their system will show you which combination of text, graphics and layouts is getting the most people to do what you want them to do.
A recent article at DM News described how Optimost performed a test of new registrations for iWon.com—using variables such as top image, signup copy, section header copy and promo copy block, and assigning values for each one.
From six variables, the Optimost system came up with 48,384 possible permutations and narrowed those down to 37 variations “based on best experimental design to maximize value of data.”
Then Optimost went into gear, watching hundreds of thousands of site visitors become members—or not.
As a result, registrations increased an aggregate of 24.1% over the initial, pre-Optimost measurement.
What were the important variables? “Section header copy accounted for 10.1 percent, the top image accounted for 6.8 percent and promo copy block accounted for 5.2 percent. Other factors contributed the rest.”
Can you assume that these relative values will be true for you? No—your mileage will vary, hence the need for a service like Optimost. Sure, you can come up with rules for building the perfect landing page, or you can simply have the machine find out the reality for you.
Turns out we still live in interesting times.
Continue reading "Let the Computer Decide" ... Read the full article