Because There’s a Little Mad Scientist in All of Us: Tips on How to Use Google Website Optimizer
| Share |
Testing out different variations of your website is a valuable tool to see what copy/offers work at getting you conversions and which ones don’t. Google’s Website Optimizer tool allows you to test almost anything on your website that you could possibly want.
Google refers to the tests the Website Optimizer runs as experiments (insert your own crash of lightning and “Mwah ha ha” here.) And, as with all good experiments, there are some parameters that you’ll need to establish before you begin:
1) Don’t change your site while running an experiment:
Just like in science class, if you have an experiment, you need a control group to base your results against. The original version of your site is your control that you’re testing new versions against. Make any changes to it, and your results will be skewed.
2) Know what kind of test you want to run:
All good scientists know what theory they’re testing before they begin. If you know what results you’re looking for, then you’ll know what kind of test to run.
If you want to test one specific thing about your site, then you would run an A/B test that tests one page against one or more alternate pages. Since A/B experiments use two completely separate pages, you can test your graphics and layouts as well as copy. You could have entirely different layout and navigation.
Resist the urge to change everything at once – copy, colors, images, offers – or you won’t know what element, or combination thereof, actually resulted in the increase in conversions. You can run an experiment where you test several different pages, each with one difference from the original.
If you want know which combination of things (headline, image, call-to-action, etc.) works best, then you would run a Multivariate (MVT) test, which tests different versions of the same page. You can have endless combinations (headline A with image B and call-to-action C, headline B with image C and call-to-action A, headline C with image A and call-to-action B…getting the picture?)
3) Know how many people you want to test on:
There’s a reason experiments are tried out on focus groups instead of the general population. That way, if something goes horribly wrong, they haven’t decimated the human race. You can use this in your own experiments.
Google’s Website Optimizer allows you to decide which percentage of people you want to run the tests on. If you don’t want to take the chance of experimenting on your entire audience and potentially losing sales if the changes don’t work, then start out by testing on 10% of your traffic.
For more, in-depth information on how to run experiments and the different variations of A/B and MVT tests, you can go to the Official Google Website Optimizer’s Blog. They have a new ebook, The Techie Guide to Google Website Optimizer, that’s free to download.
The Techie Guide is exactly what it claims to be: a tech-heavy guide that can run you through how to set up everything from simple A/B tests to the more complicated experiments with dynamic content.
**Warning** The Techie Guide assumes that you have some basic knowledge, such as how to edit HTML, the basics of JavaScript, and how cookies work. Make sure you’ve got this info down before you read it, or you might get lost when they start talking in code. (That’s not a metaphor, either, I mean actual programming code.)
Thanks for the article Cow. I haven’t yet delved into testing my sites yet, as most of my sites thus far have been “internet marketing” affiliate promotions, and after the first couple weeks or so, seem to really die down as far as the traffic goes. But by then, Im promoting the next thing. Gets old actually, and im thinking about switching to promoting something that might earn for a longer period of time.
Talk soon.
-Shaun Cartwright
Its better to know about the what is the task which we are going to develop so that it can help us in generating new and effective results.
Still Google website optimizer is out of my understanding….
Live testing is something I need to study more infact anylitics generally will be back to read more on this post, this ones a follower!
This is very useful – I will be redesigning several of my sites soon and I hope to put them through a testing phase shortly after. This article makes some great points to remember. Particularly about not making too many changes at once. Seems obvious, but so easy to overlook in one’s enthusiasm.
This is a topic that I am still having trouble with.
I use Google Analytics and I can see where traffic comes from and goes. I have wondered how to use the web optimizer.
You know what I think would be a good tool…..
A piece of software that will analyse your entire site and score the site based on some standards. The tool could scan all incoming and outgoing links (warn of broken links), image placement, text fonts, colors, etc. It is quite well established what a well designed site should look like, based on all kinds of research. Take these parameters and use as a guide for designing the “website Inspector” tool…..hmmmm maybe I should be off looking for a programmer to start this for me??
Thanks John for the great post!
Jeromy (AKA Hillbilly)
Thanks for the info Cow.
I use Analytics but have never done any A/B testing – it’s always seemed too complicated and time consuming.
(It’s only as I’m writing this that I’ve realised what a stupid excuse that is! lol)
http://w
Thanx for the information!
Very useful news. A simple example / excercise would help a lot how to do it. Anyway thanks John for the interesting post!
I just use Analytics to see where traffic soure! I love Analytics and love Google
Thank Cow!
interesing!
Well now there’s something I didn’t think of!
Thanks will give it a mowing over.
Thanks will give it a look over!
[...] Your page is on StumbleUpon [...]