SEOBy comes from "Search Engine Optimization By..." and is intended to be an informal professional organization for those that practice Search Engine Optimization, and operators of Internet search engines, vertical industry portals, and web directories

Security Expert Witness

SEO News Items - Return to News Menu

Measuring Link Popularity

by Ken McGaffin

22 Mar 2005 - LinkingMatter.com. On the face of it, measuring the link popularity of a website is simple. However, inherent flaws in the measurement process may at best give some meaningless results and at worst underestimate the effectiveness of your link building work.

Most search engines allow you to identify and count the number of links coming back into your site. On Google, for example, just enter link:www.yoursite.com and you’ll find a list of sites that link to you. You could also use one of the link checking sites like LinkPopularity that tests several search engines at once or download free software from CheckLink to easily check your own site, your clients’ and your competitors sites. However, the results from such tests should be taken with a large pinch of salt.

Let’s take an example based on some work we’ve done for a large pharmaceutical company that wanted to build on the number of inbound links they had from health portals. A quick check on their link popularity showed that they have a healthy 5,500 inbound links.

But it turns out that these links are not all they seem.

By analysing a significant sample of these links we found:

• 63% were internal corporate links
• 14% were dull, pharmaceutical directories
• 12% were academic sites
• 5% were suppliers.

We found that less than 1% - that’s an estimated 55 sites - were the type of valuable links they were after.

Now suppose we undertook a linking strategy and generated 50 additional quality links from exactly the sort of health portals they were after.

On a pure numbers basis, our work would have increased their link popularity from 5,500 to 5,550, an insignificant 1% increase. But in reality, the strategy would have almost doubled the number of quality inbound links to their site - from 55 to 105 - a very significant impact.

There are other flaws with link popularity measurement:

• To appear in link popularity results, the site upon which the link sits must have been indexed by search engines. But search engines still index only a fraction of all the web pages available. If your link sits on this ‘invisible web’, it won’t be returned.

• Even if the site on which your link sits has been indexed, there are still problems. First, if the link was added since the site was last indexed then it won’t show up. And second, if your link is buried deep within the site, it may not appear.

• Sites with restricted access or membership only areas - often sources of rich information and links - will not be accessible to search engines.
Many link popularity tests return internal links - links from within the site itself. This shows that the content has been well-structured and optimised for search engines, but it does distort the scores on link popularity. AltaVista, for one overcomes this with the command - link:www.yoursite.com -url:www.yoursite.com.

• Search results can be inconsistent. This is because different searches may be carried out on different indexes. As Craig Silverstein of Google writes, "There are many reasons why one might see a difference in the estimated number of pages returned for the same query. It’s most likely the queries ... were sent to different Google datacenters. Depending on which datacenter finishes a query, the estimated number of results may vary."

• Furthermore, a search engine’s priority is to help people find relevant information. Helping webmasters and site owners find backward links accurately may not be near the top of their priorities.

So what should you measure?

Link popularity checkers are very useful as a guide and for research purposes they can be tremendous. However, using them as a metric to judge the effectiveness of your linking work should be treated with caution.

By all means use them to give you a rough idea, but build some solid metrics into each linking project you undertake.

Here are some things you should think about:

• Analyse the type and quality of the links that currently exist. Look for any information you can use to measure effectiveness.

• Identify the top 20, the top 50 or the top 100 sites that you would like to link to you. How many of these currently link to your site? Use that as a benchmark - so if 20 of the top 100 sites currently link to you, set yourself a target of 40.

• Check your referrer logs frequently and keep a note of referring URLs. Note which URLs drive most traffic to your site.

Measuring return on investment is a key business discipline that we all have to address. Putting some thought into what you should measure for each individual linking project will not only prove your worth, but will give you valuable insights into how you can sharpen and improve your linking strategy.