Log In
Members
Home
Add Reply
Add Topic
Topic RSS
8:10 am
June 10, 2011
Offline
QuoteHey there,
I know we've talked about split testing a few times here and there on the forum. But, I don't think using free Google tools to do it has come up yet. So, I thought I'd put it forth.
Backing up a step... In a nutshell, a great way to increase signups from your squeeze page is to selectively test variations in different page elements and integrate any changes that increase signups. The simplest method is what's called A/B testing. This is where you have your current page up and running, then you create a duplicate of that page with one element changed - say your call to action, or a graphic, or the addition of a video or whatever. (but only change one thing at a time.)
Then you split your traffic between the two versions and track it until you can determine which version is giving you more signups (converting). If it's your original page then you keep the thing you were testing the way it was, if it's your alternate version, you just update your original page with the better performing element and move on to testing something else.
Well, that seems kind of complicated I know, but Google Website Optimizer makes it pretty darned straight forward. Here is the skinny on making it happen:
http://support.google.com/webs.....e=guide.cs
It'll also do what's called 'multi-variate' testing in which you are able to test a number of elements simultaneously. Both types of testing are explained pretty well. I'm just working with A/B for now because just thinking the word 'multi-variate' makes me have to take a nap.
I've been meaning to play with this stuff for a while now. I finally got around to tracking down the details and thought I'd share. Is anybody else using this tool? Any tips or gotchas we should know about?
Cheers!
Daniel
1:04 pm
August 27, 2011
Offline
Hi Daniel,
Steve brought up the tandem of using Google Analytics and Google Website Optimizer in a thread somewhere following one of the coaching calls with John - can't quite remember where it is at the moment.
He basically introduced the topic to the forum for those who weren't sure about A/B testing - similar to what you just did above. I assume that at the very least - he's using Google Website Optimizer. I haven't used it yet but definitely plan to and this is as good a place as any to share the results with each other!
6:11 pm
June 9, 2011
Offline
Daniel Crandall said:
I'm just working with A/B for now because just thinking the word 'multi-variate' makes me have to take a nap.
ah hahahahaha! Totally.
I stick to the A/B split test as well. I would imagine if I had a team of people working for me that did nothing other than split test, then I would do multivariate testing.
What's awesome about running these split test is that something as simple as one word in your headline could significantly increase your conversion rates.
Switching my submit button from "Yes, Give Me Access!" to "Get Instant Access!" increase my conversions by 7%. If I wasn't testing, I would never have known that.
Great stuff Daniel!
6:22 pm
June 9, 2011
Offline
Mike Ippersiel said:
Hi Daniel,
Steve brought up the tandem of using Google Analytics and Google Website Optimizer in a thread somewhere following one of the coaching calls with John - can't quite remember where it is at the moment.
He basically introduced the topic to the forum for those who weren't sure about A/B testing - similar to what you just did above. I assume that at the very least - he's using Google Website Optimizer. I haven't used it yet but definitely plan to and this is as good a place as any to share the results with each other!
Mike! It's so easy to use. Just set up your experiment, plop the appropriate codes into your page headers and let er rip. In fact you don't need analytics to split-test, just optimizer. I just use analytics as a back up to measure vistors to the squeeze page compared to the confirmation page to measure conversions that way too. But optimizer gives you a nice little chart to predict your winner and compare the conversion rates.
Optimizer automatically will show each variation to every other visitor (but you have to set them to show 50% of the time for this to happen). Let us know how you make out.
6:32 pm
August 27, 2011
Offline
Hey Steve,
Trust me when I say you're preaching to the converted; would love to dive into this right now but I'm severely time-challenged these days and about to depart for a week long vacation someplace warm and sunny. Using Google Website Optimizer will be a game-changer for sure.
1:41 am
June 10, 2011
Offline
Thanks guys! Hopefully this isn't too much of a duplicate topic.
Do you guys have any experience using optimizer to split test the front page of a wordpress blog?
I'm thinking you could start by creating blog page that's identical to the front page and hide it from the menu and general public with one of the page hide plugins. Then change something on the hidden page and use it as the 'test'. Seems like that would work anyway.
Daniel
6:56 pm
June 9, 2011
Offline
Daniel,
It would depend on what the purpose of your front page is for. Is your front page a squeeze page or a content page?
I wouldn't recommend split testing a blog content page for purely SEO reasons. You don't want google spotting variations of your site's front page as being a duplicate of your other front page.
To do this without google sniffing it out, you would want to make one of the variations a no follow/ no index.
The problem is because this is both a blog page and a split test, not only will you not be able to accurately predict the winner, you won't know which to no follow/ no index.
If you no follow / no index the winner, then that means that google will not index that page or will it follow any link to the rest of your content within the site. This means that pagerank won't be passed to other optimized content on your site.
I recommend only split testing pages that have a clear conversion goal. Either an opt-in or a sale. Your content should be allowed to work as seed content for search engine ranking and overall site keyword targeting to pull in your ideal audience, so that people will be pre-inclined to opt-in to your follow up even if they sign up from the blog and not a squeeze page.
I hope that's not too confusing. 🙂
7:07 pm
June 10, 2011
Offline
Hey Steve,
No that makes total sense. My thought was that maybe it might be useful for things like testing placement of feed/twitter/like buttons and stuff like that. But it was just a brainstorm, and I definitely hadn't thought it through in terms of SEO and weather it's feasible to test those elements using regular A/B methods.
Though the way I've been thinking about split testing in general is only updating the "live" version with the winning elements rather than completely switching to the winning page. But the nofollow/noindex thing does sound like a pain in the ass to manage in a blog type of platform.
Thanks man!
Daniel
5:26 am
June 7, 2011
Offline
That's one of those tools I definitely need to spend more time with. I've always just done it with a combination of aweber stats and split test options and actual results/stats. But Google Website Optimizer is definitely supposed to be better.
Having trouble with your marketing? Wish you could have an experienced direct-to-fan marketing expert look over your actual campaigns, music, or content and offer feedback? Or perhaps you’re just looking for a little one-on-one assistance so you can ask questions that pertain to your specific goals and get a second, more experienced, perspective? Click here to book a session with me now.
Finally got around to using Google Website Optimizer. It was surprisingly easy to set up. Ive never been very good with analytics. Im pretty stoked to get some feedback with this lil monster... Ill keep you posted.
DRESSED HIS WIFE UP LIKE A HORSE FOR HIS NEW ALBUM COVER
http://www.deanfields.net
9:19 pm
June 9, 2011
Offline
My favorite thing about optimizer is that it automatically switches your pages for you. You can set it to even display each version of the page so that every other visitor sees the variation instead of the original. Then it's just driving enough traffic to get some conclusive data.
easy peasy 🙂
1 Guest(s)
