Old School SEO Tests In Action (A 2014 SEO Experiment)
Ever wonder how powerful some of the oldest SEO recommendations still are? With the birds and the bears (and a little caffeine) changing so much in SEO since 2011, I wanted to see first hand some of the results we can get from some moves like internal linking and title tag optimization. Using my own site as the proving ground, and moving quickly between tweaks and first results to try and exclude any other ancillary update or change, I decided to test some optimizations I still see recommended or used in the field. The set of competing pages I chose below don’t move very often, so I thought this might be a good group to experiment with.
Note: It’s important to understand that this is not a controlled test at all. Any single domain I’m competing against could be making some changes at the same time which would naturally skew my results. Let’s take this with a grain of salt and consider all of this directional. This is not advice, this is merely my experience and thoughts. If I get hammered on this in the comments, so help me…
Truthfully I think the tl;dr can be summed up pretty well in a single statement:
[rant] See, the results of these tests turned out as I (and probably most of you) expected. Virtually no gains on the thinnest of tests. There were very little surprises below. Yet, these still bring related recommendations all the time from lesser quality blogs – or worse, sometimes agencies and consultants.
Last week I walked into a pitch where the prospect showed me some of the projects his current neighborhood SEO company is working on. He candidly told me he didn’t know what the SEO company was doing for him (which is why he was entertaining new vendors). With the draft of this post in my head, he started sharing some of the recommendations he was given – some of which coincidentally are listed below. Others recommendations included press release links and quickly churned video production.
Now I’m not one to “negative sell” over a competitor (ie, downplay someone else’s service to promote my own), and I was extremely respectful to this vendor, but I left the meeting really frustrated for this business owner. It took everything I had to keep from blasting this vendor. The business owner is clearly the victim of lazy SEO. He was a great guy trying to run a business and relied on the company to be his SEO hero. I respectfully gave him my different opinion on tactics and strategies without truly speaking my mind. I’m still not sure I shouldn’t have been more truthful.
In case you’re wondering, none of the local services in my screenshots below is the vendor I’m reluctantly protecting. [/rant]
Updated 2-20-2014: Lia Barrad made a great point in the comments that I feel should be added here. Unfortunately I couldn’t persuade a client to allow me to display bigger data. As a result I was only limited to do the tests on our own site. The amount of traffic and testing options I had on this relatively small Greenlane site didn’t give me much opportunity to also show a lift/loss in traffic. I really wanted to share that as well, because I truly think qualified and converting traffic is way higher on the list of valuable SEO KPIs. Instead I was relegated to using garbage keywords like “Philadelphia SEO” that doesn’t bring much good traffic (I used to rank extremely well for the term and eventually abandoned it because it wasn’t worth the effort in my case).
Enjoy the test!
Exact Match Internal Anchor Text Optimization
Situation: On November 17 2013, using Chrome Incognito, my site ranks #11 for a geo-targeted keyword (see graphic below – in this case I don’t want to muddy this test by adding the keyword anywhere on this website except in the testing page).
Click images for larger view
The strongest page on my site is my homepage (which is currently ranking for the keyword above). It has a PA of 52.58, with 420 external links passing link equity from 51 external domains.
My second strongest page is my Outdated Content Finder tool. It got mentions in Moz, Search Engine Land, Search Engine Journal, and picked up from mentions at Mozcon. It has a page authority (PA) of 49.07, with 89 linking root domains, for a total of 100 external, equity passing links. There are already 40 outbound links from this page, with two being to external domains.
On my Outdated Content Finder page, there isn’t a reference to the homepage using any anchor text but “home” in the navigation.
Test: To see if I could pass better PageRank to my homepage, using an exact match anchor text, I implemented the following:
- Added a link onto my second most valuable website page (the OCF tool).
- Used Webmaster Tools to “Fetch As Google”, and submit to index (for faster crawling).
Expectation: In many cases, the Fetch As Google URL submission works really fast (I’ve seen it add a new URL in less than 10 minutes), but I’m not really expecting a jump in rank. I think because the sitewide navigation, where there’s a home link already embedded, this second link may not have much power.
Result: It took a few days, but there was a single-position gain on 11/18 (same as the new cache date). The bump went from position 11 to position 10. Nothing to hang my hat on normally, but for a page jump, I’m somewhat satisfied in this case.
To push the rankings a little higher, let’s try a partial-sitewide, exact anchor link to the same homepage.
Test: 11/19 – My blog has a different sidebar than my non-blog pages. With a widget in WordPress I can add a simple piece of copy with an exact anchor text link:
This isn’t a true fully-sitewide link, and is all one level deeper into the site (http://www.greenlaneseo.com/blog/) but for this experiment I think it’s good enough.
Expectation: I have a number of blogs with a wide variety of backlinks. I still believe sitewide links have power (though limited), and expect to possibly see another position bump.
Result: On 11/24 (6 days after the change), the keyword actually dropped two spots to position 12 (page 2). From what I can observe, no new sites have entered the set.
Since that sitewide link didn’t work too well, I reversed it. Actually, I updated it to push all the links into the Outdated Content Finder page. Maybe if we consolidate into my second most powerful page it might have a positive effect to the same target keyword.
Test: 11/24 – Updated the site-wide copy as follows:
Expectation: Truth is, I expected more from Test 2. With Test 2.1, I’m even less optimistic there will be a positive change. At the least, I’m expecting my target keyword to fall back to position #10.
Result: Apparently better than expected. Now appearing in position 9 for my target keyword since 11/27.
The domains in this set stayed relatively constant throughout this 10 day experiment. Again, I make no claim to this being the results everyone should expect, since we must consider competition, possible backend algorithm changes, and (especially since these are all SEO companies) possible changes by the websites themselves. But, my theories are as follows:
- Direct internal linking with specific anchor text still has a little bit of value, especially if you direct it through your best pages. A single ranking bump from second to first page may be larger if it was a result deeper in the rankings (my guess!). So, very little gain, and definitely a small recommendation, but there are much bigger SEO fish to fry than this change. If a client had to pay to have this change done, not sure I would ever put this top on the list.
- We’ve heard sitewide links may be scrutinized by Google, and it may be true after all with a direct keyword impact dampened. But, while keyword value may not pass directly, efficient PageRank still may. Some clever “PageRank sculpting” may still have minor value. Keyword, minor. Once again, recommendations for this kind of result won’t be moving higher on my list any time soon.
Title Tag Optimization
Situation: Thousands of SEOs, websites, and audit tools suggest these two best practices for title tags:
- Target keyword should be the first word
- Title tag must be under 70 characters
Personally, I’ve rejected this for the last 8 years. Here’s why – I believe Google is more sophisticated, and realizes the target keyword being first in the title isn’t always natural. In Google’s younger days, sure – it’s a signal they could code to capture, but I think it’s too limiting to be a signal today. It’s a usual SEO recommendation that surely Google knows about. Second, if a title tag is too long, it gets truncated. That’s not a great user experience, but I’ve never seen evidence of the truncated text not helping rank. I’ve only seen the opposite.
Test: To test this, I updated my title tag for my blog homepage on 11/29/2013. Target keyword is in the middle of the title tag. I intentionally caused the tag to truncate. This is a pretty terrible title, but suits the experiment:
On a side note, after creating this terrible title tag, I submitted to Fetch as Google. Within 60 seconds this title tag showed in an incognito search, despite an outdated “Nov 18, 2013″ date. That’s remarkable.
On 11/30 through 12/02, I’m ranking position 271 for my keyword. It seems pretty settled there. On 12/03, I have updated the title tag to this:
Expectation: I don’t think the ranking will move. I don’t think keyword position matters.
Result: On 12/7 my current rank for the keyword was still 271. On 1/4/2013, it flopped down to 284. No positive change.
I wholeheartedly believe the volatility of a change is different when a rank is in the hundreds, vs. in the tens. Let’s revise the same test on a keyword that is already ranking well. For the keyword Philadelphia SEO, my homepage page ranks 6. The title tag is Greenlane SEO – Search, Analytics, and Strategy Services Since 2005. A Philadelphia SEO Company.
Test: On 1/8 let’s see what happens if I change it to Philadelphia SEO Company – Greenlane SEO – Search, Analytics, and Strategy Services Since 2005.
Expectation: I don’t think the ranking will move. I don’t think keyword position matters in this case either.
Result: On 1/12 my current rank for the keyword was still 6. No positive change (but I’m reverting immediately – that’s a terrible title tag just for a supposed SEO value).
I don’t believe that a title needs to be under 70 characters for SEO value to take hold. As mentioned early in this post, a truncated title is not great from a marketing perspective. Surely there’s better things a user can see than an ellipsis in the SERP link, but when trimming to 70 characters is recommended in order to rank better, I call “shenanigans”.
Test: I’m not going to work too hard on this test because I’ve tested this before. On 1/8/2014, on a blog post called Review of Repost.us, I rank #1 for “review of repost.” The title tag is simply Review of Repost.us. I am changing the title to past 70 characters: Review of Repost.us – A Review By Bill Sebald – Is Repost.Us SEO Friendly? Let’s Find Out! Greenlane Search Marketing
Expectation: I’m expecting no drop in rank whatsoever.
Result: On 1/13/2014, no drop with new ugly truncated title tag.
As expected, tweaking the title tags with these old-school recommendations didn’t do anything. It’s not 2007 anymore.
- Changing the location of the keyword, and extending past the 70 characters, did not seem to matter.
I do hope you enjoyed the tests. As stated in the beginning of the post, this is not scientific. Take this as directional and do what you may with the information, but my recommendation for those who still solely rely on these kinds of recommendations to provide your client with SEO services, please reconsider recommending things that have a bigger impact. If you’re a business person yourself, and you get recommendations like this, please don’t drink the kool-aid.