Working with Google Website Optimizer
I apologize for not posting in a very long time - for anyone that has noticed - but as said by the great AC/DC, "I'm Back"do do doo do "in Black"; and not wanting to disappoint, I have compiled a short and succinct entry.
**********************************************************
Google’s latest release has been in the beta pipeline for almost six months - Website Optimizer, Google's free multivariate testing application, allows an advertiser to run rather intensive landing page tests without having to recruit a lot of technical resourcing or statisticians to evaluate the data.
In fact, the interface is so easy to work with it takes most of the “test design” work out of you hands. Really, all you have to do is upload some java code to the landing pages and run through the Google wizard – in as little as ten minutes you can run your own multivariate testing (as complicated as you want to make it) without having to pay an SEM firm huge chunks of money to do it for you. Good for you, bad for me.
At SES NY, Google run a lot of insiders through a mock demo, sort of an inside pitch… what was the key take away? “It isn’t as complicated as it sounds, and the best way to get results is to just get in there and start working with it.”
Mind you, I’ve been building testing and research platforms for user interactions with sponsored and organic Search results for quite some time – not tooting any horns – so I kind of know my way around these woods already. But the truth of the matter is, if you follow a few pretty basic steps from the start, you can really start to use this tool to build some very dramatic improvements to conversion rates, page views, and visitor loyalty. So if you are in the B2B marketplace, B2C, an e-retailer, or an online publisher – Website Optimizer can help you reach your unique business goals – from meeting inventory demands to generating more warm leads…
Get it Right from the Start
1. Decide on which landing page to test. You can run tests on multiple landing pages (however, you are limiting to testing only 8 areas per page), but start off slow. Use the first test as a learning stage as well, don’t be highly restrictive, but don’t think that you can test and solve everything right away. The more complex your testing framework, and the more possible variations your run, the longer it will take to get the volume of clicks necessary to reach a statistical significance.
If you are an experienced marketer, you probably have multiple landing pages running; unique pages for product offerings, target niche, or matched user intent (if you don’t, you should!) – pick one, the best one, the one with the most opportunity for improvement for your first test.
How do you know what page that is? Look at your analytics (everyday) and it won’t take long to see which page isn’t performing as well as you think it should. But also make sure that this page gets enough click volume – you don’t want the test to take forever to give you results!
2. Brainstorm all the things you want to test. Don’t restrict your test to just running different Headers or introductory paragraphs. Try moving calls-to-action. Try making conversion paths more intuitive. Try playing with imagery that offers more resonance to your target audience. If the possibilities seem endless it is because they are.
If you don’t know where to start, take a look at your competition – what are they doing? Can you do it better? Is there something you should copy? Is there something you should avoid? I guarantee that if you spend an hour looking at what everybody else is doing, you will have hundreds of possible testing opportunities.
The trick is to try to narrow down the possible testing options. Take a good look at your conversion path and any market profiling you have done. Look to test those factors that you think will have the best results against your unique business goals. In other words, if you are an e-retailer, will testing the cart call-to-action be better than testing where you put the “contact us” link?
3. Identify the conversion page! This page is often not the landing page – if it is, I strongly suggest you rethink your strategy. The conversion page is the page your visitor gets to once they’ve taken the action you want them to take on your test page. Think “check-out” not “add-to-cart”.
4. Put the tagging in place.
"The Google Website Optimizer will generate little bits of code, called “tags,” for the Test Page and the Conversion Page. All you or your technical person needs to do is a little copying and pasting. Google will even check (validate) that you’ve placed the tags correctly before you can begin. On the Test Page, put the “Control Script” code at the top part of the page (called the “head”) and the “Tracker Script” at the bottom of the page (before the tag). You only need to tag the Conversion Page at the bottom with the “Tracker Script.” Please note the tracking script specifies in its code which is the test page and which the goal page is."
Optimizer will also give you what is called “section script” or tags that you put around those elements of the page that you want to test (remember step 2?). For example, if you wanted to test the headline of the landing page, your code would look like this:
Where “Description” would be the title of your test.
5. After coding all the testing variants – it isn’t as complicated as it sounds – hav
e Google validate the codes. They will tell you if you have done it right or if there are any errors; more so, Google will also tell you how long it will take to get statistically significant results.
6. Sit back and wait for the results… actually, no! Take this time to start thinking about your next round of testing, and testing on other landing pages. Sitting on your laurels will only cause you to take two steps back to the one forward that the testing is giving you.
There are some Caveats
In order to use Website Optimizer, well you have to have both an AdWords account and Google Analytics on your website. Both are free to set up, but a lot of marketers do not have Google’s analytics package on their website – don’t worry, having it will not affect any other analytics you may have running, but don’t expect SiteCatalyst or WebTrends to give you the same sponsored tracking if you are running a multivariate test… it will play with your reporting – but the mild inconvenience is worth it for the potential of exponential gains in sponsored originated conversions. Save to: Digg | | Yahoo | Furl | Reddit
**********************************************************
Google’s latest release has been in the beta pipeline for almost six months - Website Optimizer, Google's free multivariate testing application, allows an advertiser to run rather intensive landing page tests without having to recruit a lot of technical resourcing or statisticians to evaluate the data.
In fact, the interface is so easy to work with it takes most of the “test design” work out of you hands. Really, all you have to do is upload some java code to the landing pages and run through the Google wizard – in as little as ten minutes you can run your own multivariate testing (as complicated as you want to make it) without having to pay an SEM firm huge chunks of money to do it for you. Good for you, bad for me.
At SES NY, Google run a lot of insiders through a mock demo, sort of an inside pitch… what was the key take away? “It isn’t as complicated as it sounds, and the best way to get results is to just get in there and start working with it.”
Mind you, I’ve been building testing and research platforms for user interactions with sponsored and organic Search results for quite some time – not tooting any horns – so I kind of know my way around these woods already. But the truth of the matter is, if you follow a few pretty basic steps from the start, you can really start to use this tool to build some very dramatic improvements to conversion rates, page views, and visitor loyalty. So if you are in the B2B marketplace, B2C, an e-retailer, or an online publisher – Website Optimizer can help you reach your unique business goals – from meeting inventory demands to generating more warm leads…
Get it Right from the Start
1. Decide on which landing page to test. You can run tests on multiple landing pages (however, you are limiting to testing only 8 areas per page), but start off slow. Use the first test as a learning stage as well, don’t be highly restrictive, but don’t think that you can test and solve everything right away. The more complex your testing framework, and the more possible variations your run, the longer it will take to get the volume of clicks necessary to reach a statistical significance.
If you are an experienced marketer, you probably have multiple landing pages running; unique pages for product offerings, target niche, or matched user intent (if you don’t, you should!) – pick one, the best one, the one with the most opportunity for improvement for your first test.
How do you know what page that is? Look at your analytics (everyday) and it won’t take long to see which page isn’t performing as well as you think it should. But also make sure that this page gets enough click volume – you don’t want the test to take forever to give you results!
2. Brainstorm all the things you want to test. Don’t restrict your test to just running different Headers or introductory paragraphs. Try moving calls-to-action. Try making conversion paths more intuitive. Try playing with imagery that offers more resonance to your target audience. If the possibilities seem endless it is because they are.
If you don’t know where to start, take a look at your competition – what are they doing? Can you do it better? Is there something you should copy? Is there something you should avoid? I guarantee that if you spend an hour looking at what everybody else is doing, you will have hundreds of possible testing opportunities.
The trick is to try to narrow down the possible testing options. Take a good look at your conversion path and any market profiling you have done. Look to test those factors that you think will have the best results against your unique business goals. In other words, if you are an e-retailer, will testing the cart call-to-action be better than testing where you put the “contact us” link?
3. Identify the conversion page! This page is often not the landing page – if it is, I strongly suggest you rethink your strategy. The conversion page is the page your visitor gets to once they’ve taken the action you want them to take on your test page. Think “check-out” not “add-to-cart”.
4. Put the tagging in place.
"The Google Website Optimizer will generate little bits of code, called “tags,” for the Test Page and the Conversion Page. All you or your technical person needs to do is a little copying and pasting. Google will even check (validate) that you’ve placed the tags correctly before you can begin. On the Test Page, put the “Control Script” code at the top part of the page (called the “head”) and the “Tracker Script” at the bottom of the page (before the tag). You only need to tag the Conversion Page at the bottom with the “Tracker Script.” Please note the tracking script specifies in its code which is the test page and which the goal page is."
Optimizer will also give you what is called “section script” or tags that you put around those elements of the page that you want to test (remember step 2?). For example, if you wanted to test the headline of the landing page, your code would look like this:
Where “Description” would be the title of your test.
5. After coding all the testing variants – it isn’t as complicated as it sounds – hav
e Google validate the codes. They will tell you if you have done it right or if there are any errors; more so, Google will also tell you how long it will take to get statistically significant results.6. Sit back and wait for the results… actually, no! Take this time to start thinking about your next round of testing, and testing on other landing pages. Sitting on your laurels will only cause you to take two steps back to the one forward that the testing is giving you.
There are some Caveats
In order to use Website Optimizer, well you have to have both an AdWords account and Google Analytics on your website. Both are free to set up, but a lot of marketers do not have Google’s analytics package on their website – don’t worry, having it will not affect any other analytics you may have running, but don’t expect SiteCatalyst or WebTrends to give you the same sponsored tracking if you are running a multivariate test… it will play with your reporting – but the mild inconvenience is worth it for the potential of exponential gains in sponsored originated conversions. Save to: Digg | | Yahoo | Furl | Reddit
posted by ricktobin at
2:17 PM
|
2 comments
