Manual Penalty Removal & Recovery - Part 3 - How to Perform a Link Audit

Link auditing is a skilled process through which there is no “one size fits all” solution. It’s a fluid process in which different approaches can be adopted, depending on individual circumstances such as link history, the type of link penalty applied and the level of severity.

In the previous two posts of this series I discussed how to identify the different types of penalty applied to a website and the various factors surrounding what constitutes a link penalty. Having a thorough understanding of these two subjects will assist with the identification of any potentially harmful links throughout the link audit which may have contributed towards the link penalty.

Now that you have a thorough understand of the prerequisites, you’re ready for the next steps. By now you may be wondering:

1. How to you source the link data?
2. What are the key considerations for link scoring?
3. How do you audit the data?

In this post I will aim to cover these points and hopefully raise awareness of some of the key considerations.

Automated vs Manual Link Auditing

Before we delve any deeper into the link auditing process, it’s important to understand the differences between two types of link auditing - automated and manual.

There are tools and services on the market which have the capability to automatically categorise link data into groups of varying levels of link severity, in theory telling you which links to remove. This may seem like an appealing approach to short-cut your way through a link audit, however automation is not an approach which I would recommend.

Automated link auditing is often prone to error and often leads to misinformation therefore taking this on face value and acting upon it (through removal and disavow) can potentially inflict negative impact on organic search ranking. In turn this may lead to additional efforts and additional expense later down the line to correct the errors, by which time it may be too late. That’s not to say that the information provided through automated link classification cannot be helpful for link auditing (more on that later!) but I’d advise not relying on automation alone.

There is only one way forward in my opinion – manual link analysis and that’s the focus of this series. Manual analysis is the only way to achieve a thorough and accurate link audit with certainty over the link scoring process.

I’ll talk more on automated link classification shortly.

Understand Your Link Acquisition History

Before you delve into the link audit I would recommend, where possible, brushing up on your knowledge of any historical link building activity which may have been performed by you, your company or by a third party. This knowledge may prove invaluable when auditing each link/linking root domain.

Let’s take a quick look again at the types of links which may breach search quality guidelines:

• Advertorials
• Article Marketing
• Blog Commenting & Forum Profile Linking
• Bulk Web Directory Submissions
• Footer Links & Theme Links
• Guest Posting
• Paid Links
• Press Releases
• Reciprocal Links & Link Exchanges
• Widgets & Infographics

Previous participation in any of the methods listed above may have contributed towards manual action. Therefore by understanding your link building history you can begin to understand the reasons behind the penalty.

Try to obtain as much data as possible to widen your understanding. If possible try to include:

• The anchor text/keyword targeted
• Types of links acquired
• The source of the link
• The volume of links acquired

You can then use this data to your advantage. Cross-referencing this data with traffic and ranking data may provide an indication as to which activities, if any, may have instigated the penalty, thus setting the foundations for a more effective link audit.

Of course understanding previous link building efforts is not always feasible. Perhaps you haven’t even knowingly participated in any link building schemed, perhaps the activity occurred before you joined the company or perhaps negative SEO is to blame? (let’s not open that can of worms!) So if you don’t have access to all the data don’t fear as next we’ll aim to collect as much link data as possible.

Harvesting Link Data

The sourcing of link data is integral to performing a thorough and successful link audit. The aim of the data harvesting process is to identify and combine as much link data as possible to ensure you have maximum data clarity ahead of the audit. Failure to gather the necessary link data may mean you unwittingly omit and therefore overlook harmful links which are likely to result in a failed reconsideration request, so it’s important to be thorough.

Naturally Google recommend obtaining a list of ‘Links to Your Site’ via Google Webmaster Tools. It’s straight from the horse’s mouth so that will be sufficient, right? Wrong! Contrary to popular belief Google does not divulge every link known to them, therefore relying on this data alone often is not enough. So who do you to turn to for accurate link data?

There are many third party services which can provide extensive link data. Popular services include Ahrefs, Majestic SEO, Open Site Explorer, Blekko, Link Research Tools, and Cognitive SEO to name but a few; and let’s not forget Bing Webmaster Tools which may divulge important link data.

Some of these services are free to use, some provide limited data for free whilst others require a paid subscription. Some of these services provide a greater range of metrics than others and some provide greater accuracy than others. It’d probably be overkill (and expensive!) to use all of these services but it’s important to harness enough data to perform a thorough audit.

Each tool will inevitably provide the data in their own format therefore once you’ve acquired the necessary link data from your chosen sources chances are you will have to slice and dice the data into a coherent format. With this in mind it’s important to ensure you have the necessary link metrics at your disposal.

Link Metrics & Data Intelligence

Some of the popular tools used for sourcing link data will provide a broader range of metrics than others. Google for example provide only the source URL and the date of link discovery via Webmaster Tools whereas Bing will provide the source URL, the destination URL and the anchor text of each link. It’s therefore important to understand which metrics will prove beneficial to the audit.

At a bare minimum you should have access to the following metrics for each link:

• The source URL – the URL on which the link resides
• The destination URL – the URL on your site which the link points to
• The link anchor text

Most data sourcing tools will provide these metrics as they are widely associated with link evaluation and commonly used in link auditing; however if you want to be a bit more scientific with your link auditing, it’s often beneficial to have the following additional information:

• Clarification of whether the link is contains the ‘nofollow’ attribute
• Whether the link is still active or not
• The IP address of the link
• Domain registrant data

These additional metrics will allow for a more extensive analysis of each link providing the ability to group and segment the data into different levels of categorisation.

Automated Link Classification

Many of the popular link intelligence tools such as Link Risk, Link Detox and Cognitive SEO also have the ability to provide the automatic classification of links. In my earlier introduction I touched briefly on the dangers of relying on automatic classification but, when used correctly these additional metrics can prove beneficial.

In my experience, automated link classification is usually presented in two forms:

• The classification of link type (directory, article, press release etc)
• The classification of link severity (a scale of harmless to harmful)

Whilst these commercial tools undoubtedly have sophisticated methods of link classification they remain a result of algorithmic software calculations and are therefore can sometimes be unintentionally misleading.

“We only give you opinions and recommendations. We ask you to review and double check the automated results. Make your personal decisions for any actions based on your own experience and direct viewing of the nature and content of websites displayed as potentially harmful to your backlink profile and website. The links evaluated by the program as harmful do not always affect your search engine ranking.” – Link Detox

If you are using link intelligence tools, by all means use the link classification metrics as a guide. The classification methods of software such as Cognitive SEO or Link Detox are extremely intuitive, algorithmically assessing each link/domain against a number of low-quality measurements to ascertain quality vs risk. More often than not such software packages also allow you to manually override the default classification based on your own analysis.

So use the data to your advantage but don’t necessarily rely on the default classification and do not let these classifications cloud your judgement on the severity of each link.

If you’re working with a large quantity of link data then such classification methods may provide a starting point for data segmentation but will ultimately require manual analysis if you are to perform an accurate audit.

Auditing the Data

We’ve now reached the tricky bit, the actually auditing of the link data. There is no “one size fits all” solution. Link auditing will require the manual assessment of each link within your link profile and will require a level of expertise to ascertain the quality of each link.

I aim to provide guidance towards those link signals often perceived as unnatural but ultimately you’re going to need to be intelligent with your reasoning and adopt your own method of link scoring. If you have any doubts about a link or linking root domain remember Chris’ golden rule: play it safe!

Do not ring fence any links for removal unless you are absolutely sure of the potential implications.

Link Scoring - Considerations

So let’s once again recap on the link building practices often deemed as unnatural.

• Advertorials
• Article Marketing
• Blog Commenting & Forum Profile Linking
• Bulk Web Directory Submissions
• Footer Links & Theme Links
• Guest Posting
• Paid Links
• Press Releases
• Reciprocal Links & Link Exchanges
• Widgets & Infographics

Again I would urge you to bear in mind that not all links in the above categories are unnatural so you’ll have to be intelligent with your link scoring. For example, a link from an informative, well written, intelligent guest post which resides on an authoritative industry relevant website will likely be perceived as credible, where as a link from a poorly written, uninformative guest post on a ‘jack of all trades’ style blog is likely to lack credibility; however if the aforementioned informative guest post utilises exact-match anchor text(s) then Google may perceive the link as unnatural. It’ll inevitably be down to you to decide which links are in breach of search quality guidelines and which are not.

With this in mind you may wish to adopt several levels of link scoring, for example:

• Links which require removal
• Linking root domains which require removal
• Links which are acceptable
• Links which reside on authoritative domains but may require a change of anchor text
• Link which require addition of the nofollow attribute

You may wish to expand on this list but these scoring methods will provide a great starting point. You may even wish to assign different levels of severity for those links requiring removal i.e. those requiring immediate removal, those which have borderline severity, and those which have potential to become a problem in the future.

Single Link Removal vs Domain Level Removal

The disavow process itself inevitably takes place after the link audit but it is first important to understand the key differences between link and domain disavowing.

In a nutshell there are two disavow commands:

1. Individual links
2. Entire domains

Disavowing individual links asks Google to ignore that particular link when assessing your website where as disavowing a domain asks Google to ignore all links from that domain.

It is important to understand and consider the difference between the two when performing your audit. It is possible to negatively impact the organic search reach of a website by unintentionally disavowing an entire domain rather than a single URL from that domain. This will require a strong understanding of link credibility and link scoring metrics. And remember the golden rule, if in doubt, play it safe!

For more information on the types of disavow command I would recommend a quick read of this post from Google Webmaster Central.

Auditing Methodology

The actually method of link auditing will again depend on your own individual requirements. Each audit is unique and whilst each may follow a similar process, each inevitably requires its own bespoke approach.

• If you believe you have had targeted manual action against a single keyword or a group of keywords then you may wish to filter your link data by anchor text and work through those exact-match anchors first.
• You may wish to filter by link classification (using classification type guide as a guide only!) and work your way through groups of similar link types.
• You may wish to filter the link data by IP address and look for those within similar IP ranges or those on bad link neighbourhoods.
• You may wish to filter the data by domain registrant using this information as an indicator to help identify link networks
• Or you may wish to simply sort the list A-Z and work your way though it each link.

Whether you adopt one of these methods or whether you adopt a your own approach the bottom line is you’re going to have to manually score each link if you are to be thorough with your efforts. So find a method which works for you.

Link Scoring - Unnatural Link Signals

Now we come to the most important part - the manual scoring of each link. Below are some of the considerations which you should bear in mind when scoring each link within your link profile but this is by no means an exhaustive list. More importantly the scoring of each link may vary depending on the size of your link profile and the history link acquisition.

1. Overuse of broad match or exact match commercial anchor text (i.e. keyword driven)
2. Links which reside on websites generally deemed as on low-quality and/or outside of website guidelines, including those mentioned previously:
   • Advertorials
   • Article Marketing
   • Blog Commenting & Forum Profile Linking
   • Bulk Web Directory Submissions
   • Footer Links & Theme Links
   • Guest Posting
   • Paid Links
   • Press Releases
   • Reciprocal Links & Link Exchanges
   • Widgets & Infographics
3. Links which reside on pages utilising poor-quality content, for example:
   • Poor use of language
   • Poor readability
   • Scraped content
   • Spun content
   • Thin content
4. Links which reside on domains de-indexed by Google
5. Links residing on hacked websites
6. Links residing on adult websites
7. Links which reside on seemingly authoritative yet untrustworthy websites
8. Links stemming from domains with minimal topical relevance
9. Links from any other “bad neighbourhood” spam sites
10. Links which reside within a link network
11. Links which reside only on the homepages of referring sites
12. Non-language specific links appearing on sites with content in a foreign language
13. Site-wide links
14. Redirected domains to specific money-making pages

I’ll re-emphasise once again that if you are unsure for whatever reason then play it safe and if necessary seek expert help. As explained previously, not all links which fall into the categories above will be malicious or harmful, therefore be cautious and thorough with your efforts.

Conclusion

So there you have it, the basic concepts and methodologies behind link auditing. It’s not ground breaking and it’s not rocket science but it does require a level of expertise to ascertain the severity of each link.

If you follow these methods the results should be a comprehensive link audit with accurate scoring of each link. It’s important to remember that the audit forms the basis of the link removal and the link scoring will be extracted to produce the disavow file, through which it is entirely possible to significantly harm an organic search profile if the link scoring is not accurate.

With this in mind I would urge you to ensure you are cautious with your auditing efforts and if you’re in any way unsure then enlist the help of an expert!

In the next post I’ll look at when you should create and submit your disavow file before moving onto the final step of the process: filing a reconsideration request.

If you have any questions or queries please get in touch and I’ll be happy to help.