Anchor Text Variation in your Link Profile: Do It
Hi all, my name is Emma and you may remember me from such B3Labs posts as How to identify kill-stealing in your online campaigns and Judging a website by its Domain Authority Profile. A recent post of mine focused on one element of a natural link profile, so I thought I’d do a follow-up on another aspect: anchor text variation.
I’ll start by saying that there are quite a lot of good blog posts out there on this subject; in fact, Tim Grice at SEO Wizz wrote one not long ago. I thought I would follow up the idea that ‘exact anchor text has had its day’ with some research and messing around with Excel.
For this case study, I used six websites – three ‘winners’ and three ‘losers’ in SEO. These come from Search Metrics organic visibility charts. Below you can see the most recent (as of 28/08/2012) charts for relative winners and losers in the UK.
Here are the processes I went through – you can follow this for whichever website you want, although I’d recommend doing it for more than one website, as then you can compare and contrast.
- I chose three winners and three losers in different industries
- Pulled their link profiles from my favourite website using the filters below:
- I then converted the Anchor Text tab into lower case by using the formula: =lower(Anchor Text) where ‘Anchor Text’ is the cell reference for the Anchor Text column.
- I created a new column called ‘Keyword Type’
Here, we are going to separate brand keywords from non-brand keywords. Brand keywords are keywords which include the brand name or a variation of the brand name. For example, for a brand like Marks and Spencer, the ‘brand’ terms would include the words ‘marks and spencer’, ‘marksandspencer.com’, ‘M & S’, ‘M&S’, the list goes on…
There are a few ways of doing this:
1. Manually (this takes forever but is the most accurate)
2. Conditional Formatting on the ‘Anchor Text’ column (good, but requires sorting and filtering and all that jazz). To do this, choose ‘Conditional Formatting, Highlight Cell Rules, Text which contains’ and then enter the relevant brand rules. You then change your range to a table (‘format as table’), sort the Anchor Text column by fill colour, and then add ‘brand’ to the filled cells in the ‘keyword type’ column and the rest are ‘non-brand’. This may have to be applied more than once if the website has several brand terms.
3. An IF Formula (requires brain power). The formula I used was =IF(ISERR(FIND(“brand”,Anchor Text,1)),”NON-BRAND”,”BRAND”) where ‘brand’ is the relevant brand term (in lowercase) and ‘Anchor Text’ is the cell reference we used before in the lower formula. Nested IF formulas may have to be used if there is more than one brand term.
Convert to a table if you haven’t already and make sure you have pasted the values of your ‘Keyword Type’ column into the column otherwise any formulas you may have used might break Excel. You should end up with something like this:
Create a Pivot Table (under ‘design, summarise as pivot table’) with ‘Keyword Type’ and ‘Anchor Text’ as your row labels and ‘Count of Anchor Text’ your values. You should then have this nice little table which you can expand:
Separate the brand keyword from the non-brand keyword, sort it out however you like, make it pretty and then compare to other websites!
Below shows what I found. The three tables at the left-hand side show the length of the keyword variation tables for the ‘SEO Winners’ and the three right-hand side one are the length of the ‘SEO losers’ keyword variation. There is a noticeable difference in the length of these – that is, the ‘Winners’ tables are always longer – this means they have far more keyword variation than the SEO Losers. Therefore, varying your keywords is something you should look into if you want to have a natural looking profile!
Apologies for my horrible Microsoft Paint pictures, I’m sure the web design team could make it look nicer, but that’s beside the point.
Recall that the ‘winners’ and ‘losers’ came from Search Metrics, and were not just deemed ‘winners’ and ‘losers’ by myself. It seems clear that in both sets of results, the anchor text variation (number of rows) is much greater in the ‘winners’ tables than the ‘losers’ tables.
However, you must take into consideration that the greater number of links you have, the naturally greater the anchor text variation should become. So it is important to take into consideration the weighting number of links each website has if you are going to do a comparison like this. Maybe use a random sample of the links that Open Site Explorer generated, so that when you run this analysis you can more proportionally see the size of the anchor text variation tables.
You can do this by using the =rand() function in Excel and then sorting that column any way you like. Once the column is sorted, picking the first however many rows will result in a random selection.
Essentially, we have seen that anchor text variation is something that is common among the SEO ‘winners’, but why is this? Think about it this way – how do people normally link to a website? Do they all use anchor text rich links, and do they all spell your brand name the same way? Look back at the links in this article.
I’m pretty sure no one else has linked to SEOWizz using the anchor text “Tim Grice at SEO Wizz wrote one” and does the term “my favourite website” tell you (or rather, Google) much about what Open Site Explorer is? Not really! If you’ve read a lot of blog posts, you know that people tend to link from really random sentences and phrases, so Google probably sees it a bit suspect if your website has only 3-4 different keyword variations out there.