In the search marketing industry, we like to believe we’ve got hard and fast rules as to what goes into proper SEO. From the way we choose keywords to the length of page titles, to how often and where best to place links from one page to another and across web sites. The fact however, is that we don’t have direct access to the search engine algorithms, so we need to rely upon our experiences and those of others in the industry we trust. An aspect that we think applies today may not tomorrow. And how that aspect applies may be open for debate, like the Meta Description has, at times, or at least it has, until now?
Last week, I posted an article in response to a claim I read where it was reported that Google is now using Meta Descriptions in page relevance. We got a flood of input, both on Twitter and in the comments of that article. Vanessa Fox chimed in, as did Jill Whalen, Edward Lewis, and about a half dozen others. So a few people agreed to join in and test this out…
When it comes to testing things we think might be ranking factors, it’s not possible to test as completely or thoroughly as we’d like simply because we don’t have access to the search engine algorithms, nor do we have the capacity to test under all circumstances without gaining access to the server resources that companies like Google or Microsoft have.
Given that reality, we need to make assumptions and break the scientific rule by going with our “best guess”. In my life, I prefer to call this following the intuitive. Most of you probably call it going on instinct, or “a gut feeling”. Then, as more data comes in, or as I can further study what happens after I’ve taken action based on that, I can sometimes gain more knowledge.
The bottom line here is this – don’t take anything I offer here in terms of recommendations as if they were facts. Always be sure to do your own testing. And if you find results different than mine, don’t be afraid to speak up.
Test One – Brand New Page
I started out by creating a brand new page on my older business site – a site that had previously had nine pages indexed but where I had recently set up 301 redirects on all but the home page, since I am no longer offering those services.
For this test, the new page had a Title of “Google Meta Description Usage Test”, and a Meta Description of “This is a Meta Description Usage page set up to see whether Google is messing with Meta Descriptions or not…”
The page was then seeded with full and partial variations of the page Title, however I made sure that the chunk “Google is messing with Meta Descriptions” was not repeated in the content.
I then provided a link to the new page from the home page, and a link to it within my original blog article about this issue.
Finally, I added the new page to my sitemap.xml file and resubmitted that to Google through the GWT system.
Within an hour or so the new page had been indexed and was coming up as it should for the primary phrase.
Unique Meta Description Not Found
I then conducted a search for the exact match on “Google is messing with Meta Descriptions” and Google couldn’t find any match.
Test Two – Existing Page – Not Unique Text
My old business web site home page contains the phrase-chunk “to clients throughout Marin and the bay area”. A search at Google produced seven results using an exact match search, none of which is that home page. Six of the results that came up did include that phrase in their content area – and all are sites that list my site along with the full Meta Description. The only other result was for a bay area builder that has that phrase in their content as well, none of which have it in their title. My page did not come up in the results.
Test Three- Existing Page, Unique Text
For this test, I chose a site where I’ve got little control over the content – there’s minimal content at best on the page I used. And there’s a unique phrase in the Meta Description nowhere else on the web. This one is a client’s site with more traffic than even my blog (which isn’t hard to attain, believe me!)
Once again, We’re not seeing any results in the SERP for that exact phrase.
I’d share other tests with you that I conducted, but the fact is that they all got the same results.
Other people’s Supporting Data
Other tests were conducted by some great minds in the community in conjunction with my tests.
So it would seem this whole thing started with someone mis-quoting Maile. Which isn’t uncommon, as was pointed out by Vanessa Fox back when my earlier article was discussed, as backed up by Dawn Wentzell who herself was at that SES keynote that started this whole debate. Yet it was Vanessa’s initial response that she wasn’t aware of Google using the Meta Description as a relevance factor, that prompted Jill Whalen to chime in with just the opposite view. If you’re interested in that longer discussion, check out the comment thread over on that original article. I’d post them here but that article ended up with 49 comments :-)
The bottom line, as of the date of this article, is that while there is a belief by some people that the Meta Description IS used as a signal to this day, so far, none of the tests conducted around the globe in conjunction with this research have been able to verify that. Other tests may be ongoing, as others had said they wanted to join in, and if anyone comes up with different results, I’ll update this article…
For me however, it will be business as usual – I will continue to recommend Meta Descriptions be well thought out, include at least that page’s most important keyword phrase, and be written in a way that communicates in human readable format, that THIS is the result in your search that you were looking for. Even if it turns out that someone can validate the usage (in a documented way that can be replicated), I have a feeling I won’t need to change my methodology much, if at all.