Nine analyses you can perform to determine whether your link profile is ‘natural’

Wesley van der Hoop
9 augustus 2016
Terug naar blogoverzicht

Nine analyses you can perform to determine whether your link profile is ‘natural’

According to several studies, backlinks are one of the most important ranking factors when it comes to organic search engine ranking. There used to be a time all you needed to do, was to obtain a high amount of links from high authority websites. No matter what that link looked like. Those times are (almost) over.

It has become absolutely critical to take into account the quality of the links referring to your website. Google algorithm updates like Penguin are getting better in recognizing so called ‘unnatural’ link profiles, and will downgrade the rankings of the website accordingly.

This leaves us with the question: how does a natural link profile look like? And are the websites we are working on, doing a good job? Underneath, you will find nine analyses you can perform to gain insights in what a natural link profile looks like.

As an example, a study of the link profile of the Dutch consumer electronic industry is being provided. We included the 22 leading domains in this market, analyzing over 35,159 backlinks.

 

 

Link profiling: Black-hat vs White-hat

A link profile analysis can be used for both good and evil. In an ideal world you wouldn’t need to analyze a link profile to make sure your linkbuilding activities are in line with the Google guidelines. When you are solely using white hat linkbuilding tactics, the link profile should be fine.

However, often a link profile already has been established, before you enter the company. Also, a link profile analysis is a great way to find out whether a third party linkbuilding partner, is doing a good job.

To be sure there are no blackhat tactics being used, it might be good to analyze the current link profile, compared to the most organically visible websites in the same industry.

According to the outcome of your analysis, it might be appropriate to shift linkbuilding tactics, or to take more rigorous measures - like disavowing spammy links to clean up your link profile.

 

 

First things first: Do affiliate links contribute to your link profile?

Affiliate links are part of a link profile, but are not contributing to the authority of the page or domain. Affiliate links are by definition paid links. According to the Google webmaster guidelines, all paid links should be provided with a ‘no-follow’ attribute.

However, affiliate links can also be filtered out by the Google algorithm automatically. The algorithm is able to recognize affiliate links of the major networks (Matt Cutts, 2012).

Affiliate links can often be recognized by the parameters included in the URL. To determine what a natural link profile looks like, it is important to filter out these affiliate links first.

 

How to do it?

Export the link profile to a spreadsheet. This can be done by a multitude of tools, of which Majestic, Ahrefs and MOZ are the most commonly used tools in the business.

Filter out all URL’s containing a parameter, and find out whether the parameters are linked to an affiliate network. In the spreadsheet, then exclude all links containing affiliate parameters.

Take for example the following URL: www.website.com/product?AffilliateID=123

In this case you want to exclude all URL’s containing ‘?AffilliateID=’.

Now all affiliate links are removed from your data, it is time to start analyzing the remaining backlinks!

 

 

 

Analysis #1: Follow versus No-follow links

No-follow links are mostly paid links (advertorials) and links that are placed in user-generated content (blog post comments, forums, reviews). In the past, these links were often abused to boost the rankings in a unnatural way.

Although no-follow links do not contribute much to the authority of the website in most cases, they are often an important part of a natural link profile. An unnatural amount of no-follow links might be a sign of black hat linkbuilding techniques. When the proportion of no-follow links is way higher than that of your direct competitors, it might be worth to further investigate how those links came into existence.

 

How to do it?

Export the link profile to a spreadsheet and calculate the ratio between follow and no-follow links.

 

How it looks like in the Dutch consumer electronics industry:

 

Follow vs Nofollow-backlinks

 

Since no-follow links contribute very little to the authority of the website, they will be filtered out after this analysis. The remaining links will be ready for further investigation.

 

 

 

Analysis #2: Page authority distribution

The authority of the page which contains the external link to your website, is one of the most important factors when it comes to the valuation of a link. The higher the page authority of the page on which the external link is located, the more likely it is that the link contributes positively to the authority of your website.

However, when your website has a link profile in which the page authority distribution is not at all correlated with successful other websites in the same industry, this might be a sign of over-optimization.

Note: For this study we have used MOZ’s page authority. However, Ahrefs (Ahrefs Rank), Majestic (Citation Flow) and other tools have similar scores to measure the authority of a webpage.

 

How to do it?

Export the link profile to a spreadsheet and exclude affiliate and no-follow links. Then categorize the links based on page authority.

For example: count all links with a page authority from 1 to 10, 11 to 20, 21 to 30 etc.

Once you make a graph based on this data, it will provide you the insights you need to compare the distributions of various websites.

 

How it looks like in the Dutch consumer electronics industry:

 

 Page authority distribution

 

 

 

Analysis #3: Homepage links versus Deeplinks

Deeplinks are the external links which refer to any page on the website, in exception of the homepage. Most link profiles have both homepage links ánd deeplinks. A high amount of deeplinks might be a sign of over-optimization. For instance, building too many links directly to commercially attractive pages, such as category- and product pages.

 

How to do it?

Export the link profile to a spreadsheet and exclude affiliate and no-follow links. Then count all external links referring to the homepage of the website.

 

How it looks like in the Dutch consumer electronics industry:

 

Homepage links vs Deeplinks

 

Note: Websites with a high brand awareness often have a higher percentage of homepage links, than websites with low brand awareness. This can be explained by the fact that the brandname will be mentioned on the internet, more often than websites with low brand awareness.

 

 

 

Analysis #4: Anchor text distribution

After measuring the page authority distribution, the percentage of no-follow links and the percentage of deeplinks, it is time to dive a bit deeper into the anchor text. The anchor text is basically the text a user has to click on to follow the link. For image links, the anchor text is determined by the alt description of that image.

There is no difference between the value of a text link and an image link, as long as the image contains an optimized alt attribute.

Anchor texts can be categorized in six different categories:

 

All anchor texts containing the brandname of the website. Eg. Netprofiler (example, not an actual link).

Anchor texts with the web address of the page in it. Eg. www.netprofiler.nl.

All anchor texts that – on itself – do not describe the page it is referring to at all. Eg. Click here or Go to the website.

All images that are linking to the website, containing an alt attribute.

All images that are linking to the website, missing an alt attribute.

Anchor texts that are not branded and describe the webpage it is referring to. This category often contains keyword optimized anchor texts. Eg. Digital marketing agency or Cheap televisions.

 

As you might have noticed, it is possible an anchor text can possibly fall into more than one category. Take for instance an image link with the alt description ‘Netprofiler - Digital marketing agency’. You can categorize this anchor text as ‘branded’ (Netprofiler), ‘image link with an alt attribute’ ánd as a ‘describing anchor text’ (Digital marketing agency).

In this case, you can choose to make a seventh category, containing all anchor texts with both the brandname ánd the product or service provided. You can also choose to just mark it as ‘branded’. Whatever you do, do it consistently over all the websites you are analyzing.

A below average ratio of branded anchor texts and a disproportionally high percentage of ‘describing anchor texts’ might be a sign of over-optimization. When your profile shows many image links without an alt attribute, there might be an opportunity to fix those ones as well.

 

How to do it?

Export the link profile to a spreadsheet and exclude affiliate and no-follow links. Then start categorizing the links based on the anchor text.

 

 

How it looks like in the Dutch consumer electronics industry:

 

Anchor text distribution 

 

 

 

Analysis #5: TLD distribution

TLD stands for ‘Top Level Domain’, which is basically the last part of the URL of the root domain (eg. .com, .co.uk, .org).

 

TLD explained

 

Top level domains can be divided into various categories. Most common are:

 

 

For a more extensive list of TLD’s, click here (actual link with generic anchor text).

 

To visualize the link profile, in most circumstances it is advised to group all TLD’s in three categories:

 

 

When the percentage of ccTLD’s of the marketplace is relatively low, this might be a sign of weak local signals.

It is recommended to take a separate look on the ccTLD’s that do not represent the countries the website serves. Combine those ccTLD’s with a spamscore check, to see whether those links are part of blackhat linkbuilding tactics. Be especially cautious with TLD’s that tend to have a reputation to be misused for blackhat linkbuilding purposes (like .ru and .in – amongst others).

 

How to do it?

Export the link profile to a spreadsheet and exclude affiliate and no-follow links. Then start categorizing the links based on the TLD.

 

How it looks like in the Dutch consumer electronics industry:

 

TLD distribution

 

Note: ICANN-era gTLD’s are not a ranking factor for search engines. It doesn’t boost the rankings to have a keyword inserted into the TLD (source).

 

 

 

Analysis #6: Percentage of link directories

Adding links to link directories used to be one of the most popular linkbuilding tactics. Often link directories have a high page- and domain authority, and content doesn’t have to be created to gain a link on those websites. Basically, it resulted in low effort, high impact backlinks.

Times have changed though, directory linkbuilding has become a lot less effective and might even have a negative effect on the link profile of a website in some cases. When doing a link profile analysis it might be useful to check the quality and percentage of link directories, compared to the (strongest) competitors in the industry.

 

How to do it?

Export the link profile to a spreadsheet and exclude affiliate and no-follow links. Then start categorizing the links based on root domain.

Make an extensive list of root domains that are known to be link directories, and mark all matching root domains as ‘link directory’.

Note: It is quite extensive labor to put the list of link directories together. I found over 600 root domains in the Netherlands alone and the list is still growing. Any suggestions of how to filter out link directories fully automatic, are very very welcome.

In the Dutch consumer electronics industry, we found that approximately 25% of the backlinks came from link directories.

 

 

 

Analysis #7: Relevancy

Relevancy has become a lot more important than it used to be. Ideally, the subject of the page ánd the domain linking out, should be correlated to the service of the website it is linking to.

 

How to do it?

This is a tricky one. Ideally, an extensive list of keywords that can be considered ‘relevant’ is compiled (use a keyword research report, if it is already been made).

In an ideal situation you then scrape the content of every page that is linking to the website. Partly, this can be done with crawl software like Screaming Frog.

This way, you can weigh the ratio of relevant keywords used in:

 

 

For link profiles containing a large amount of backlinks, it might be hard to automate this process. Normally I would just check the list of relevant keywords with the pagetitle of the linking page. Measured this way, the amount of ‘relevant links’ is shockingly low (around 2%). I expect this to increase once modern linkbuilding methods will gain more traction, focusing on getting more relevant links.

 

 

 

Analysis #8: Spamscore distribution

The spamscore is added quite recently to MOZ and other tools with link monitoring capabilities. This score consists out of several checkpoints that might indicate low quality of the link. The higher the spamscore, the more likely it is the link can be considered low quality.

It is advised to double-check the links with a spamscore higher than three. Almost every link profile has a few questionable links. Although these links should always be kept to a minimum, the website should be fine unless the amount of spammy links is disproportional high compared to other websites in the same industry.

I wouldn’t advise to disavow all links with a somewhat spammy profile. Unless you already have been penalized by Google, those links still might add some value to your website’s authority. Just try to avoid spammy links as much as possible in the future.

 

How to do it?

Export the link profile to a spreadsheet and exclude affiliate and no-follow links. Then start categorizing the links based on spam score.

 

How it looks like in the Dutch consumer electronics industry:

 

Spamscore distribution

 

 

 

Analysis #9: Link velocity

Link velocity is the rate at which new links are won – or lost – over a certain period of time. Link velocity itself is not a direct factor when it comes to building up authority. However, you might want to investigate sudden increase or decrease in the amount of backlinks. Investigate the quality of the links won/lost in that short period of time.

When the amount of links increases over a short period of time, and the quality of those links is low, maybe some black hat tactics have been used - possibly by a third party linkbuilding partner, or even internally.

When the amount of links increases over a short period of time, and the quality of those links is high, evaluate how the links were gained, to replicate the same circumstances in the future.

The same accounts for competitors: when a website in your industry gained a tremendous amount of high quality backlinks in a short period of time, it might be worth to further investigate the methods used – if any.

 

 

 

Bringing it all together

Repeat analysis #1 to #9 for all websites in the same industry. Once all the data is gathered, it is time to compile it all together. The strongest competitors with the most organic visibility normally would set the bar for other, less visible websites.

 

How to do it?

Export all the outcomes of the nine analysis above in a spreadsheet, and add the organic visibility of the websites. Now calculate the weighted average, based on that organic visibility.

 

How it looks like in the Dutch consumer electronics industry:

In this case we compared the link profile of website X with the weighted average of the twenty most visible websites in the market. The data outcomes seems to give an indication that the link profile of website X is not naturally obtained.

 

Page authority distribution of a natural linkprofile

The page authority distribution seems to be way off limits compared to the other twenty websites in the same industry. This might be an indication of over-optimization. Linkbuilding efforts are possibly focusing too much on high authority pages.

 

Spamscore distribution of a natural linkprofile

The spam distribution indicates a high amount of spammy links are obtained in the past. It might be worth investigating those links.

 

Anchor text distribution of a natural linkprofile

 

On top of that, compared to the weighted industry average, website X has a high amount of describing anchor texts (19.23% compared to an average of 9.74% for other websites in the industry). When zooming in on those terms, the describing anchor texts seemed to be mostly focused on commercially attractive keywords.

All these signals together make the link profile of website X suspicious to say the least.

 

Note: Since we are working with percentages, make sure the link profile has enough followed linking root domains to gain trustworthy insights of the linkbuilding activities.

 

 

Conclusion

Ideally you wouldn’t need to analyze link profiles at all. However, most of the times there already have been some linkbuilding activities before you start working on a website. Data of competitive link profiles in the industry can give insights of how a natural link profile looks like.

This being said, a link profile analysis is a great way to define the borders. It is not wise to push the link profile of a website towards these borders.

When your analysis tells you the amount of spammy links of the industry is around 20%, and yours is 15%, this doesn’t mean you want to build more spammy links. It does indicate, however, that you are doing well keeping the spammy links out – compared to your competitors.

Define the borders of a natural link profile, but try to improve on quality when building your own link profile.

 

Deel deze blogpost
Neem contact op met Wesley van der Hoop

Was deze post interessant voor je? Lees meer nieuws van deze auteur. Voor meer informatie kun je ook bellen of mailen.

mail Wesley van der Hoop 020 345 9488
Meer lezen van Wesley van der Hoop
Reacties (0)
Reageren