SEO Friendly Split Testing

I’ve recently been spending most of my time split testing the designs across many of the sites I manage. The question is always raised – how will this effect my seo? Unfortunately website owners aren’t allowed to serve up different page versions to search engines than to regular visitors – without risking the wrath of Google. (That is unless you’re one of the big brands) This technique is commonly referred to as “cloaking” and is expressly in violation of Google’s webmaster guidelines.

Luckily we have a new tool at our disposal to set things straight with some level of reliability – the link canonical tag. According to Google:

If your site has identical or vastly similar content that’s accessible through multiple URLs, this format provides you with more control over the URL returned in search results. It also helps to make sure that properties such as link popularity are consolidated to your preferred version.

Let’s assume three split test variant pages:

  • index.php
  • indexb.php
  • indexc.php

To properly point each page to the original you would simply place this tag in the HEAD statement of each page:

<link rel="canonical" href="">

What about other search engines you say? Well luckily Yahoo! and Bing announced support for the link canonical tag at the same time as Google. YMMV with these second-tier search engines – but at least they have stated their intent to honor it.

A second level of protection can be added by adding a meta tag to each page to prevent indexation. I would recommend a setting of NOINDEX, FOLLOW as seen below:

<meta name="robots" content="NOINDEX, FOLLOW" >

Obviously this method is ONLY going to work if you are split testing and not multivariate testing.

Firefox Keyword Research Addon

If you are anything like me you probably use seobook’s keyword rankings tool enough to have it on speed dial. Luckily the script used to process results accepts a GET request. As a result I have made a simple firefox search engine addon to make it easier to get the keyword research you are looking for.

NOTE: I am not in any way affiliated with seobook or Aaron Wall. (Aaron if you’d like me to take this down send me a note – Ryan)

xml for search engine:

<ShortName>seobook keyword research</ShortName>
<Description>seobook keyword research
<Image height="16" width="16" type="image/x-icon"></Image>
<Url type="text/html" method="get" template="{searchTerms}"/>

To install this script just click the down arrow on your search engine bar in firefox and select “seobook keyword research”. Big thanks to Aaron Wall for creating such a useful tool.

PHP & MySQL Keyword Rankings Script

I’ve been working on a Google keyword rankings script that scrapes Google search engine results pages and stores the values in a mysql database. The bulk of the heavy lifting is done via php & cURL. I used some jquery to allow manual updating of rankings. Much of the scraper code was lifted from an old 5ubliminal post that I can’t seem to find anymore. The following video is a short clip of a portion of the script in action. You can test out the script here.



  • connect.php – stores credentials for mysql connection
  • add-domains.php – allows you to add a domain into the database. This domain is then presented as an option on the index page.
  • add-keywords.php – checks a domain (already inserted via add-domains.php) against a keyword. If a match is found it inserts the data into the database.
  • index.php – displays all of your domains
  • final.php – the actual keyword rankings page. Allows you to update rankings and see each domain’s keyword rankings on one page.
  • style.css – some css styling
  • postjax2.php – processes serp results and stores values in the db
  • keyword_rankings.sql – database scheme – load this query via phpmyadmin to setup your mysql db
  • header.php – some navigation
  • ajaxproxy.php – some cross domain proxy I found to allow you to get rankings from multiple ip addresses

Google doesn’t particularly like it’s serps scraped – sometimes they will return garbage data if they suspect you are doing it. To get around this, we must bounce our requests off of multiple IP addresses. If you have a few cheap hosting accounts lying around, upload the postajax.php script to a folder accessible from the webserver. Insert your urls where these files can be found into the following section of code:

   var csurla = 'http://YOUR_URL_HERE/postjax2.php';
   var csurlb = 'http://YOUR_URL_HERE/postjax2.php';
   function cointoss(wot){
         var d= new Date().getSeconds();
         wot=(wot)? wot: [csurla, csurlb];
         d+= Math.round(Math.random()*10);
         return wot[d%2];

The above example randomly chooses one of two urls to post the data to. This code could easily be changed to accommodate as many urls as you have access to. The next segment of script gathers the data from each table cell in a row of data. We then serialize the data and post it to to our ajax proxy which gets around the pesky cross domain ajax restriction. the $.Ajax command that is built in to jQuery then waits for the php proxy script to respond and then inserts the resulting data into the row we just clicked:

$(".update").click(function() {
       /* extract all the info out of the table cells */
       var selectR = $(this);
       var urlsearch = $(".urlse a").html();
       var se = $(selectR).find("").html();
       var wt = $(selectR).find("td.wt").html();
       var str = $(selectR).find("td.keywords a").html();
       var ranking = $(selectR).find("td.rankingspan").html();
       var orank = $(selectR).find("td.orank").html();
       var delta = $(selectR).find("td.hclass div").html();
       var clicks = $('#tabler').find("td.clickstotal").html();
       var wttotal = $('#tabler').find("td.wttotal").html();
       var keysearch = str.replace('/%20', '+');
       var csurl = cointoss();
       var dataString = 'csurl=' + csurl + '&clicks=' + clicks + '&wttotal=' + wttotal + '&domain=' + urlsearch + '&keyword=' + keysearch + '&wt=' + wt + '&se=' + se + '&rank=' + ranking + '&orank=' + orank + '&delta=' + delta;
       $(this).html("<td colspan='9' align='center'><img src='http://PATH_TO/loading.gif'></td>");

          type: "POST",
          url: "http://PATH_TO/ajaxproxy.php",
          data: dataString,
          cache: false,
          success: function(data){
          var totalr = 0;
          var tempint = 0;
          if($(this).text().length > 0){
           var tempint = parseFloat($(this).text()); 
           totalr += tempint;

       if($(this).text().length > 0){
           var tempwt = parseFloat($(this).text()); //convert to number
           totalwt += tempwt;
           totalwt = totalwt.toFixed(2);