Tuesday, March 10, 2009 3/10/2009 09:39:00 AM
We'd like to share with you two implementations of Google Website Optimizer.First, a fellow named Bud Gibson. Bud runs a non-profit called Michigan Innovators, focused on connecting innovating companies with innovative people. Bud put together a test to try to improve monetization of the job board on Michigan Innovators' site.
Bud goes into great detail about how he designed his experiment, including a thorough explanation of WiderFunnel's (a GWO consultant) LIFT method. His test is on track to increase his conversions by 25%. Congrats on the success, Bud!
Next, our friends at the Conversion Rate Experts (also GWO authorized consultants), have begun working with Vodafone. They haven't gotten results in, as their experiments have just begun, but it's fantastic to see a large company like Vodafone taking website testing seriously.
Two very different companies using website testing to improve their businesses. A nice example of how Google Website Optimizer can work for you, no matter your business.
7 comments:
Trevor, thanks for the write up regarding the experiment we are running at Michigan Innovators. One thing I have discovered in running this experiment is that it can take some time for the experiment to settle down. Right now, I've had a high confidence result (over 98% probability of beating the original) for the past couple of days. The lift is good but about half the 25% it looked like after the first week.
What I think has happened is that I am getting some strong day of week and time of days effects. One of the combinations is beating the original by a giant margin during working hours on the week days but staying just even with it on the weekends. I also think I have some different populations of job seekers hitting the site.
I'll have more to say in my write-up on the results in a week or two. Future efforts with Google Analytics and Google Website Optimizer will try to tease these out.
Truth be told, the insight one gains into a site's visitors from tools like GA and GWO is as valuable as the lift you get out of any one experiment.
BTW, it's clear to me you must be calculating a dispersion parameter for the variance and s.e. in GWO. Are there any write-ups on the underlying statistical algorithms you're using.
Can Google Website Optimizer work for me ?
I,m Just a small business Owner I dont have all that kind money these Big companies!
Certainly. Keep in mind that if you don't have very much traffic, you'll want to keep your experiments to just a couple of variations, but Website Optimizer can be used on any website.
It's one of google's exciting feature. I am enjoying on it.
I have used Google Website Optimizer for 2 experiments and I feel that it's very slow. I have started the experiment about 2/3 months ago but still it's not showing any report rather than prompting not high confidence.
How the Google Website Optimizer's work can be speed up?
Hi Pabitra,
There are two factors that probably count the most towards duration of your experiment. The first is the amount of traffic you have. On average, you need at least 100 conversions per variation to confidently show a winner.
The second factor is differentiation. If you have two variations and one converts at 10% and the other converts at 10.1%, you need a lot of data to prove that the second variation is actually better. On the other hand, if that second variation converts at 20%, then you won't need as much data to show a winner.
Google website optimizer is a great tool. It is very useful.
Post a Comment
You are welcome to comment here, but your remarks should be relevant to the conversation. We reserve the right to remove off-topic comments, or self-promoting URLs, and vacuous messages. Please keep your comments relevant, friendly, and respectful.
We moderate all comments so you may not see your comment show right away.