Wednesday, 6 February 2013

Anchor Text Optimization Case Study: What’s Natural?


A significant component of the Penguin update appears to deal with anchor text over-optimization of backlinks. Savvy SEOs have been concerned with maintaining a natural anchor text profile for years now, and the Penguin update has just underscored how important natural anchor text optimization really is.
What constitutes a natural anchor text profile? The reality is that a natural mix of anchor texts is far less optimized than you might think. Geoff Kenyon made a post on SEOmoz suggesting that 65% non-targeted, 35% targeted is a good mix to aim for.
I believe that this is such an important issue that I wanted to do my own study to add a few (post-Penguin) data points to the conversation. In this post, I’ll share the results of a quick study I completed on 9 websites. Each website chosen for this analysis has a strong brand and is ranking well for an important keyword.
Notes about the analysis:
  • Anchor text data was pulled from Open Site Explorer.
  • Anchor texts from all links pointing to the root domain were used for calculations. (I also looked at data for only the ranked page; in most cases there wasn’t a big difference.)
  • Number of root domains linking was used for calculations.
  • Due to the way I filtered results, domain and URL anchor texts were generally lumped together as branded anchor texts.
  • In general, I did not count product names, product lines, etc. as branded terms. If they were counted, the percentage of branded anchor texts would increase.
  • Unique anchor text instances is calculated by adding the “Number of Linking Root Domains Containing Anchor Text” column in the OSE cvs export.
  • Primary goals were to estimate natural usage of branded anchor texts, and to take a major keyword the site ranked for and analyze how much that keyword was used in the anchor text profile.

Hunter Sailboats

www.huntermarine.com is ranked #1 on Google for “sailboats”
I analyzed 769 unique anchor text instances:
  • Branded anchor texts (contains text “hunter”): 559 linking domains, 73% of total
  • Exact Match anchor texts (for keyword “sailboats”): 0
  • Phrase match anchor texts (contains phrase “sailboats”): 20 linking domains, 2.6% of total
  • Partial and broad match (contains “sailboat”): 23 linking domains, 2.9% of total (this is due to “boat” being part of the brand name)
  • Of the top 20 anchor texts in use, all are branded

Ranger Boats

www.rangerboats.com is ranked #1 on Google for “bass boats”
I analyzed 1,938 unique anchor text instances:
  • Branded anchor texts (contains text “ranger”): 1,139 linking domains, 59% of total
  • Exact Match anchor texts (for keyword “bass boats”): 3
  • Phrase match anchor texts (contains phrase “bass boats”): 13 linking domains, 0.7% of total
  • Partial and broad match (contains “bass” and/or “boat”): 876 linking domains, 45% of total (this is due to “boat” being part of the brand name)
  • Anchor texts including “bass”: 25 linking domains, 1.2% of total
  • Website has a significant amount of pharmacy-related links. Assume that a black hat SEO got content onto the domain and tried to get it ranked.

Mercury Marine
www.mercurymarine.com is ranked #2 on Google for “outboard motors”
I analyzed 6,091 unique anchor text instances:
  • Branded anchor texts (contains text “mercury”): 3,315 linking domains, 54% of total
  • Exact Match anchor texts (for keyword “outboard motors”): 1
  • Phrase match anchor texts (contains phrase “outboard motors”): 34 linking domains, 0.5% of total
  • Partial and broad match (contains “outboard” and/or “motor”): 543 linking domains, 9% of total
  • Of the top 20 anchor texts in use, the only non-branded one is “click here”

EZ Loader

www.ezloader.com is ranked #1 on Google for “boat trailers”
I analyzed 402 unique anchor text instances:
  • Branded anchor texts (contains text “ez”): 215 linking domains, 53% of total
  • Exact Match anchor texts (for keyword “boat trailers”): 1
  • Phrase match anchor texts (contains phrase “boat trailers”): 53 linking domains, 13% of total
  • Partial and broad match (contains “boat” and/or “trailer”): 147 linking domains, 37% of total
  • Of the top 20 anchor texts in use, the non-branded ones are: roller trailers, pontoon trailers, aluminum trailers, bunk trailers, parts guide, request a brochure, manufacturer’s recall clearance letter, personal watercraft, parts e store, adjustable trailers, adjustable brochure

Lund Boats

www.lundboats.com is ranked #1 on Google for “fishing boats”
I analyzed 765 unique anchor text instances:
  • Branded anchor texts (contains text “lund”): 545 linking domains, 71% of total
  • Exact Match anchor texts (for keyword “fishing boats”): 0
  • Phrase match anchor texts (contains phrase “fishing boats”): 7 linking domains, 0.9% of total
  • Partial and broad match (contains “fishing” and/or “boat”): 383 linking domains, 50% of total (this is due to “boat” being part of the brand name)
  • Anchor texts including “fishing”: 43 linking domains, 5.6% of total
  • Of the top 20 anchor texts in use, the only non-branded ones are “website” and “more info”

What About eCommerce?

The sites above that I analyzed are manufacturers with strong brands. So what about e-commerce sites? Will their backlink profiles have a lower percentage of branded anchor text? (After all, they have a lot more products that can be linked to, and webmasters are as likely to be linking to a specific product on the site as to the site/brand itself.) I checked a few to find out:
Conclusion: E-commerce sites anchor text profiles are still primarily branded.

Conclusions

  1. The majority of links analyzed (about 55-75%) included the brand/website in the anchor text.
  2. The most common natural anchor texts are variations of the brand name and website URL.
  3. Exact match anchor texts were very rare (apparently they are not always needed to rank, even for competitive  keywords).
  4. Phrase match anchor texts are also uncommon. Most anchor texts that included keywords were partial match anchor texts.
While this study is certainly not comprehensive, it does serve to give a snapshot into what natural anchor text distribution looks like for a website with a strong brand. Please don’t take the numbers I’ve presented here as industry standards, but I do hope these numbers will help serve as a catalyst to encourage SEO’s to look at anchor text optimization a bit closer.
The data from my study matches relatively nicely with the data from Geoff’s study referenced earlier in this post. My data does show more branded anchor texts and lower keyword anchor texts. It’s hard to tell if this could be Penguin related, due to differences in the markets we analyzed, or simply because both our data sets were pretty small.

What does a natural anchor text profile look like?

The data from this study suggests that:
A natural link profile will rely most strongly on branded anchor texts (some of which may also contain keywords), secondarily on broad/partial match keyword anchor texts and other anchor texts (i.e. click here, website, etc.), with only a very rare exact match keyword anchor text thrown in.
Based on the data from my study and Geoff’s study, I believe the following may be a useful rule of thumb:
  • 60% brand, URL, brand+keyword, and non-targeted anchor texts
  • 30% partial, phrase, and broad match keyword anchor texts
  • 10% exact match anchor texts
I would love to hear from other SEO’s who have done their own analysis or have a different perspective on the data. What are your conclusions regarding anchor text optimization?
 Tip: To get more accurate numbers for your site, analyze competitors in your industry that you know are relying on natural linking.



Read Source: http://www.searchenginejournal.com/anchor-text-optimization-case-study-whats-natural/43404/

No comments:

Post a Comment