DaveP Posted August 30, 2012 Share Posted August 30, 2012 Apologies in advance if this is old news, but for anyone who hasn't realised how trivially easy PW makes this, just give it a try. Google Analytics (GA) has a newish feature - 'Content Experiments' (multivariate testing if you want to be posh), and this is so easy to do with PW, it's almost unbelievable. What it boils down to is testing different variations of a page to see which performs (gets more sales, signups, leads, whatever) better, either between 2 versions (A/B testing) or more (multivariate testing). Say you have a page with a download link (or 'sign up for newsletter' or whatever), and you want to test size, colour, position or whatever of the download link - Install Ryan's Page Clone module Clone your page as many times as necessary, so you end up with/download-page/ (the original) /download-page-1/ /download-page-2/ etc [*]Edit page titles so they are the same as the original [*]Set the copies to hidden, so they don't appear in navigation, search, etc., but are still accessible by their URL [*]Change whatever in each copy (link colour, button size/colour) [*]Create a Goal in GA (I'm not a GA expert by any means, so you're on your own with that bit, but it is quite easy) [*]Create a new experiment under Content > Experiments and point it at your original and variations [*]GA gives you a bit of JS to add after the <head> tag of the original page, which allows them to redirect to the variations as necessary, so you need this in that page's template. [*]Sit back and bask in how easy it was and await GA doing the hard work Eventually, GA will detect which variation performs better, and you can end the experiment and dump the other page variants, replacing the original with the winner. 4 Link to comment Share on other sites More sharing options...
Pete Posted August 30, 2012 Share Posted August 30, 2012 Does GA do the work of loading one of the variations when the original page is requested then? If so then that's pretty cool! Link to comment Share on other sites More sharing options...
DaveP Posted August 30, 2012 Author Share Posted August 30, 2012 Pete, yes it does. I missed a bit in the original post, which I'll add now. (Forgot the bit about the JS snippet to allow them to do all the redirecting etc) Link to comment Share on other sites More sharing options...
Pete Posted August 30, 2012 Share Posted August 30, 2012 This is awesome. I always thought of it as testing two options with massively different layouts, but it makes sense that just increasing a button size or making a slight tweak on one design will allow you to refine an already good design over time. I'm definitely interested in testing this out, and of course you're right that it's made really simple with ProcessWire - clone a page, change its template to a slightly altered one (or just some page content even) and you're away! Of course, the more visitors your site gets the better the test results will be - it would be difficult for a brand new site that only gets a handful of visitors a week, so I think it's more for sites that are already established. Link to comment Share on other sites More sharing options...
apeisa Posted August 30, 2012 Share Posted August 30, 2012 Pete: handy also if you are planning to throw lots of money to Google for paid visitors. You do want to get best possible conversion then. Link to comment Share on other sites More sharing options...
dennisgorelik Posted August 30, 2012 Share Posted August 30, 2012 Are you sure Google Analytics Content Experiments support Multivariate testing already? Lack of Multivariate testing was mentioned multiple times as the biggest problem of Google Analytics Content Experiments (in comparison with Google Website Optimizer that Google killed way too early). Please don't confuse A/B testing with Multivariate testing. Apologies in advance if this is old news, but for anyone who hasn't realised how trivially easy PW makes this, just give it a try. Google Analytics (GA) has a newish feature - 'Content Experiments' (multivariate testing if you want to be posh), and this is so easy to do with PW, it's almost unbelievable. What it boils down to is testing different variations of a page to see which performs (gets more sales, signups, leads, whatever) better, either between 2 versions (A/B testing) or more (multivariate testing). Say you have a page with a download link (or 'sign up for newsletter' or whatever), and you want to test size, colour, position or whatever of the download link - Install Ryan's Page Clone module Clone your page as many times as necessary, so you end up with/download-page/ (the original) /download-page-1/ /download-page-2/ etc [*]Edit page titles so they are the same as the original [*]Set the copies to hidden, so they don't appear in navigation, search, etc., but are still accessible by their URL [*]Change whatever in each copy (link colour, button size/colour) [*]Create a Goal in GA (I'm not a GA expert by any means, so you're on your own with that bit, but it is quite easy) [*]Create a new experiment under Content > Experiments and point it at your original and variations [*]GA gives you a bit of JS to add after the <head> tag of the original page, which allows them to redirect to the variations as necessary, so you need this in that page's template. [*]Sit back and bask in how easy it was and await GA doing the hard work Eventually, GA will detect which variation performs better, and you can end the experiment and dump the other page variants, replacing the original with the winner. Link to comment Share on other sites More sharing options...
DaveP Posted August 30, 2012 Author Share Posted August 30, 2012 My bad - from the Google Support page I linked above. Content Experiments is a somewhat different approach from either standard A/B or multivariate testing. Content Experiments is more A/B/N. You're not testing just two versions of a page as in A/B testing, and you're not testing various combinations of components on a single page as in multivariate testing. Instead, you are testing up to five full versions of a single page, each delivered to visitors from a separate URL. turns out it's neither. (Although, and my expertise is limited here, you could use it for either, as well.) Link to comment Share on other sites More sharing options...
Recommended Posts