One of the benefits of writing my own personal SEO-guide blog is that I can test changes and then report positive or negative results without worrying that my personal studies may damage the websites of my company or our clients at my day job. I was reminded of this principle when I recently tested an over-optimization variable here.
As most SEM professionals know, the alt-text of a photo or graphic on a website (as long as it is not in Flash) is one of the many on-page items that can be optimized for search engines through the use of keywords. As I wrote in a prior post on SEM-PDF strategy, search engines can only read the code of a website — or, in other words, text — and not images. So, in Google’s digital eyes, an image is reduced to the text that is included in the alt-text field.
Since every image on a page can have a keyword inserted in that field, I knew that this website — see the images in the left and right banners — had several “opportunities” that had not been “optimized.” From the six social-media sharing buttons in the top right-hand corner to the advertisements from SEO-software companies, I counted roughly thirteen additional places on the home page where I could place the main keyword-target of this website: “seo software” (of course).
So, out of curiosity, I added the keyword to all thirteen images:
Logically, of course, this site’s ranking for “seo software” should have skyrocketed, right? After all, the placements were thirteen additional ways to tell Google that, yes, this site is extremely relevant to people looking for SEO software, right? Well, not such much:
In my SERP spreadsheet, I list the positions of all keywords in each months’ columns as they were at the end of those months (though I actually check weekly). I made the aforementioned changes at the end of June 2011. The entries for the present month — July 2011 — are a week old at the time of publication and not finalized.
Here’s the point: As I have grown this site, the Google SERPs for my main targeted-keyword (“seo software”) increased from 200+ to 68 over seven months. Then, after I made the described changes to the alt-text, the SERP declined from 68 to 112 after a couple of days at the end of June. While a certain amount of fluctuation will always occur, the decrease in the SERP was too significant to be random. In statistical terms, it was beyond the general margin-of-error and one standard-deviation. And after I removed the changes, the SERP reverted to its original position in a couple of days in early July.
SEO Guide: What NOT to Do
So, obviously, this result had meant that Google had slapped this site with an over-optimization penalty — the keyword density for “seo software” had seemingly grown beyond a reasonable level — that had been removed once I had fixed the “problem,” right? Well, the honest answer is that it is difficult to know. The reason is complicated.
Having in mind that the recommended keyword density is from 3 to 7%, anything above this, say 10% density starts to look very much like keyword stuffing and it is likely that will not get unnoticed by search engines. A text with 10% keyword density can hardly make sense, if read by a human. Some time ago Google implemented the so called “Florida Update” and essentially imposed a penalty for pages that are keyword-stuffed and over-optimized in general.
SEO Design Solutions writes:
I have seen everything from people abusing the noscript tag to stuff links, creating 1 pixel tables and keeping the text the same color as the background (to hide content), stuffing alt text to optimizing CSS values with keyword-rich anchor text to get a boost in search engines.
The point is, things taken past the point of moderation are a red flag waiting to happen. We understand that competitive markets exist and not everybody plays fair (such as database driven dynamic insertion based on keyword clusters, RSS arbitrage, cloaking and other gray areas), but the last thing you should do is follow suit in an attempt to inflate relevance.
So, it seems pretty obvious that Google directly penalizes sites that over-optimize themselves for SEO, right? Well, Matt Cutts of the search-engine giant addressed the issue in this short YouTube video (via Search Engine Land) and keeps the issue a little murky:
The gist: While Google states that it does not assign explicit over-optimization SERP penalties per se, the fact remains that websites that use such techniques will be viewed by the public as less user-friendly at best and potential spamming at worst. (Would you want to read a website that had a certain phrase repeated in every single sentence? As a writer and former journalist, I would be disgusted.)
While Cutts seems to contradict what the prior experts have said, I would argue that everyone whom I have cited is saying the same thing but in different ways. As I wrote in a prior post on the SEO effects of social media today, search-result positions are going to be increasingly determined by factors including usability, quality, and the amount of social-media sharing rather than just keywords, links, and the use of search-optimization software. A bad site will have bad rankings.
Cutts and everyone else, in my opinion, were merely stating different parts of the overall issue:
Over-optimization —> Bad user experience —> Bad site —> Low SERPs
Cutts was correct in saying that that over-optimization does not directly lead to Google penalties and low SERPs. And the others were simplifying the issue for their target audiences of beginners by saying that A directly leads to D while there are actually a few steps in between.
SEO and SEM is both an art and a science. As much as online marketers love data, the fact remains that we are all merely trying to make educated guesses as far as what exactly Google likes — and all analysis will be imperfect as a result. As much as it is a cliche, I must describe the difference in statistics between correlation and causation:
- Causation — If I do X, then Y will occur directly as a result
- Correlation — Websites that do X also tend to have Y results (whether or not X directly causes Y)
The two terms are not identical. Say that a website has great on-site optimization, quality content, good social-media sharing along with a high rank in search engines. It can be accurately stated that websites with all three of those factors also tend to be those sites that have high rankings (the correlation is high). Since it is very difficult to determine what exactly and directly causes higher SERPs, the best that SEM experts can do is to emulate those factors that are correlated with sites that have high SERPs. (In a different context from my journalism blog, a college education is correlated with high earnings in general, but a college education does not always directly lead to higher salaries in any specific case.)
Regardless of whether over-optimization directly causes penalties and/or low search-result rankings, the fact remains that sites that use the practice seem to be correlated with those that have poor results in Google and elsewhere. So, I would avoid the tactic.
At SEJ, Jennifer van Iderstyne writes how bloggers and webmasters should approach the issue — namely, that it is a better use of time to create a quality website with unparalleled content rather than squeeze a keyword and link into every small, possible place:
I also believe that link building involves creating valuable content, building networks and participating in communities, social media, blogs and even real physical communities. These are all, in their own ways, passive forms of link building. Because site promotion, and building awareness may not result in immediate links but they create circumstances in which your content might be linked to freely in the future.
When you take the time to make your site an educational authority, or a unique destination within your niche you can and will attract links that you don’t have to ask for. And that notion brings us full circle back to over-optimization.
I completely agree.
Samuel J. Scott, a former journalist in Boston turned Internet marketer in Israel, is the founder and publisher of My SEO Software and Director of Digital Marketing and Communications and SEO Team Leader at The Cline Group. You can follow him at Google+, LinkedIn, Facebook, and Twitter. His views here and elsewhere do not necessarily reflect those of his company and clients.