Season 1, Episode 2, SEO Fight Club

<iframe width="640" height="360" src="https://www.youtube.com/embed/71ZEsNgprlA" frameborder="0" allowfullscreen></iframe>

http://seofightclub.org/episode2

Hi and thank you listening to SEOFightClub.org. I’m Ted Kubaitis and I have 22 years of web development and SEO experience. I have patented web technologies and started online businesses. I am both an engineer and a marketer. My goal is to help you win your SEO fights.

Maybe you are new to SEO or maybe you have been doing SEO for as long as I have. This SEO primer will reboot your effectiveness and point you in a direction for achieving positive results that continue to build on new learning. Most SEOs fail in the “O” part of SEO. The optimization part of SEO demands iterative learning based on empirical measurements. In this and future episodes I’m going to show you how fight the good SEO fight.

Columnist Chris Liversidge at SearchEngineLand posted an article on August 25,2015 called Mega Menus & SEO.

In this article Chris aims to answer the question “What are the best practices surrounding mega menu navigation elements when it comes to SEO?”. Under the heading of “Best Practices In Mega Menu Implementation” he lists “Code order placement” and later states “my first piece of advice for mega menus is this: Try to include it in the HTML after the main body content.”

If I took Chris’s advice then I would go to my company’s engineering team and submit a work order to re-architect the pages across 20 different websites. The costs would be rather large. I’d have to get management involved and testers. The cost of even redeploying 20 large scale websites is high too. With all the people and time I can only make 2-3 of these kinds of requests each year so I need to be sure that making this change is going to help my SEO results. If I make the company spend all this time to do all this work and the result is “no change” then I lose credibility, people will likely be mad at me for wasting their time, and there is an opportunity cost because I could have suggested other work that does matter. I need to test this idea that the Mega Menu “should” be below the main content.

This episode’s FREEBIE: http://seofightclub.org/episode2

With every episode I love to give something away of high value. This episode’s SEO freebie is the data I collected to prove my case. This data comes from software I created. The software makes empirical measurements of on page SEO factors. It also calculates the statistical correlation of each factor’s measurements versus the results ranking position in the search results. This gives me strong clues as to which factors are plausibly influencing my rankings and which are not. The software measure over 300 different factors and today I am giving you the data for all the keywords I researched for this topic. So if you are an SEO data junkie like me then this freebie is a gold mine!

First lets spot check a high competition keyword and see what the winning sites are doing. Maybe we’ll get lucky and find some clues in the data.

I guess we should start by explaining how to see if the mega menu is above the main content. I use chrome as my primary web browser. To see if the mega menu is above the main content I right click the mega menu and select “Inspect Element” the mega menu is nearly always in a wrapping DIV tag with an ID like menu,nav,navigation, or even megamenu. When you hover the div tags in the inspected source chrome highlights the contents of the the div in the browser’s HTML view. By simply hovering the divs below the menu’s wrapper div you can quick see if the mega menu is above or below the main content.

Search Term: toys

Odds are pretty good that most of these page one companies are pretty invested in professional SEO services. Given they are top ranking for a highly coveted search term I think it is safe to say they are successful at it too.

#1 Toys R Us mega menu above the MC

#2 Target.com mega menu above the MC

#3 Walmart.com mega menu above the MC

#4 Amazon.com mega menu above the MC

#5 Kmart.com mega menu above the MC

#6 Kohls.com mega menu above the MC

#7 Costco.com no mega menu

#8 Sears.com mega menu above the MC

#9 TopTenToys.com no mega menu

#10 IMDB page for movie “Toys” mega menu above the MC

So 80% of the page 1 results had a mega menu above the main content.

20% had no mega menu at all.

0% had a mega menu below the main content.

Not a great start for Chris’s hypothesis, but these initial results are actually inconclusive. Just because nobody is doing it doesn’t mean there isn’t advantage to placing the mega menu below the main content.

Clearly having you mega menu above the main content is a very common standard and it looks like Google is accepting of the practice. So at a glance having your mega menu at the top doesn’t appear to hurt you. This means our investigation can shift to explaining a more basic premise behind the hypothesis.

Does having your keywords near the top help you rank better? The idea is that having your mega menu push your keyword content further down in the source code hurts your rankings. I think we need to measure this factor and see if it correlates at all with rank position in the search results. If it does then Chris’s hypothesis will be plausible and if it doesn’t then we can all send Chris our grumpy faces for gambling with our time and resources in his advice.

My software measures a factor “Number of Bytes To The First Match In The HTML Source”… If Chris’s hypothesis is true then I would expect to see a correlation when comparing this measurement to ranking position. The result? No Correlation. The values do not trend with rank position. The values look like they are in a random order… Here they are round to nearest 100 bytes going from #1 on google to #10:

2700, 2000, 8000, 200, 20000, 300, 2100, 10000, 200

I know… I have only tested 1 keyword. What if a pattern emerges after 10 keywords. Lets try that.

Search Terms: toys,toys for boys,toys for girls,toy sale,online toy store,discount toys,cheap toys,buy toys,buy toys online,best toys

The result? No correlation… again let me share the values, but this time the values are the average number of bytes to the first match for the rank position in the search results for all the search terms in our sample. So it is the average bytes to the first match for all the number 1 results and the average number of bytes for all the number 2 results and so on.

Again going from #1 in google to #10 the averages across the whole sample set:

700, 1MB, 7000, 1000, 7000, 700, 1MB, 3000, 2000, 2400

Pretty random huh? Random means no correlation with rank position. Let me geek out on math a little here. I promise to keep it short.

A Correlation Coefficient describes how well your data points fit to a trend line. Something that is a significant factor should trend as you approach number 1 on google. As you grow your sample size the strength of that ranking signal should improve.

There are two correlation coefficients that are frequently used. Spearman’s which is typically used for things that have a curved trend line and Pearson’s which is good for things that fit a straight trend line. My software calculates both.

When we looked at toys all by itself the Spearman’s Correlation was -0.14 and the Pearson’s Correlation was 0.02. When the values are close to zero it means your measurements appear to be random with respect to rank position. The strongest correlation would have a value close to 1 or -1. So when I say the values look random and there is no correlation… that is not my opinion. That is a mathematical fact. And the statistics people listening will want to know my critical values and significance. For this study Spearman’s needs a critical value of 0.65 and Pearson’s needs 0.63 to be 95% certain that the distribution is non-random. Our correlation values are nowhere near that and that is the math telling us that these measures appear to be random.

When we switched to a ten search sample the Spearman’s correlation became -0.02 and the Pearson’s correlation became -0.17. They stayed near zero even after adding more searches to the sample. That means they look random across our sample set too.

This is pretty compelling data that the number of bytes to the first keyword match is not a ranking factor for google at this point in time. It may have been last month and it might be next month, but right now it is not. And that is the big problem with relying on your personal experiences too much. To know the truth or fact of the matter you have to constantly challenge your own beliefs.

But we are not done yet. It could be that Google discounts certain things like javascript and inline CSS and other HTML gobble-de-gook. All those thing could make out bytes to the first match appear random and boy do they appear random. So until we rule that out are study is still inconclusive.

My software measures another factor. The number of matches in the first 100 words after stripping all the HTML out of the source.

So