Then you'll want to ask him if that necessitates a separate URL. If it does, you should rel=canonical the test page to point at the original home page, and noindex the test page to avoid any duplicate content situations.

You may also want to run this by the optimizely guys to see what their thoughts are, since they'll know exactly how their technology is altering your site.

My understanding is that the same URL is used for both test and control versions of your home page, meaning its not actually redirecting. Instead the changes are inserted using the javascript code dynamically. I'm not sure whether Google ignores the javascript change or incorporates it. Either way your test changes should be done with both SEO and user experience in mind, so don't run tests that would hurt your SEO and you'll be golden.

They both point to the same location on the server? So there's not a different folder for the subdomain?

If that's the case then I suggest adding a rule to your htaccess file to 301 the subdomain back to the main domain in exactly the same way people redirect from non-www to www or vice-versa. However, you should ask why the server is configured to have a duplicate subdomain? You might just edit your apache settings to get rid of that subdomain (usually done through a cpanel interface).

Here is what your htaccess might look like:

<ifmodule mod_rewrite.c="">RewriteEngine on
  # Redirect non-www to wwww
  RewriteCond %{HTTP_HOST} !^www.mydomain.org [NC]
  RewriteRule ^(.*)$ http://www.mydomain.org/$1 [R=301,L]</ifmodule>

If you're doing on-page SEO, you can't focus on broad match, you can only optimize for the exact keyword phrases you use on your site in H1 tags, titles, and all the rest.

You could try to find all the broad match synonyms and stuff them into your title and H1 etc etc but then you're diluting the SEO value for any given page by doing so. Best to focus on 1 or 2 keywords per page and optimize for that.

My suggestion is to take the large volume you get from broad matching, and find the exact phrasing that takes the lion share of that volume and go with that.

As Dan mentions you must be careful though... within a set of related keywords that broadly match a phrase, some keywords are at the beginning of the buyers funnel and others are at the end. Make sure your site is optimized to take people along that funnel. Don't only optimize for the biggest search phrases that only people not actually buying use.

Without a doubt add it to the main site under a folder like /blog. As John Barth mentioned, using it as a subdomain will pass all the domain authority value to the subdomain and not your main site, so its functionally equivalent to hosting it on a different domain.

I assume you're talking about PPC because broad vs exact match doesn't really make sense from an SEO perspective: you can only optimize a site for the exact phrasing you use on the site and hope SE's match it with broader phrases.

Whats striking is the large difference here between the broad and exact volume of searches.

• 33,000 for broad phrase is somewhere in the middle of the curve. Its not highly trafficked, but its nothing to sneeze at either.
• 500 for exact phrase match is very low.

The difference means that the exact phrase you chose isn't what most people are searching for, even though the topic is mid-sized volume wise. You'll want to try using keyword research tools to find the phrases people are using for this topic.

In terms of getting an idea of how big the keyword is, compare it to others that you've had success with. Its all very relative because some companies will have an easy time capturing large popular keywords that a company with a smaller budget couldn't touch,

There was a trick in the Geddes adwords book that talked about finding related keywords and I can't for the life of me remember his tactic. It had to do with using operators in regular searches. Anyone remember that?