SEO · Complete guide
How to analyse a website's backlinks
A reliable backlink audit follows the real order of the work: collect, map, assess, then act. Here is the method, with its limitations and a worked fictional example.
In this article 8 sections
- 01 · Prerequisites: access, tools and the question to ask before auditing
- 02 · Step 1: collect the backlink data
- 03 · Step 2: map the link profile
- 04 · Step 3: assess the quality of the backlinks
- 05 · Step 4: spot risky links and decide what to do
- 06 · Worked example: auditing the link profile of a fictional blog
- 07 · Common mistakes to avoid
- 08 · Practical tips for a lasting audit
In brief: a backlink audit takes place in four stages: collect dated data, map the links, assess them by hand, and only then decide whether to act. Tool scores are for sorting, not for passing judgement.
The figure displayed in large type in your tracking tool is not a Google score. It is a vendor-specific metric, calculated using the vendor's own data and its own rules. This page walks through the method in the real order of the work, stating at each step what the data allows you to conclude, and what it does not. A complete fictional example closes the process.
Prerequisites: access, tools and the question to ask before auditing
Before opening a single report, two things matter: knowing what data you have, and knowing what you are looking for. Without that, you collect a lot and conclude little.
Your own site or a competitor's
There are two situations. If you manage the site, you can access a Search Console property: this is Google's own data, but only for your site. If you are studying a competitor's site, or one you do not own, only a third-party index tool will give you a view of its links.
These two sources do not overlap. One reflects what Google has retained for a site you control; the other reflects what a vendor has itself discovered on the web. Confusing the two is the first source of misinterpretation.
Next, write down an objective in a single sentence: understanding the authority acquired, spotting risks, or studying a competitor. This objective determines how deep each step goes. As for the role of links, it can be summed up in a few words: they contribute to how a site gains authority and relevance in the eyes of search engines. To place this work in a broader approach, see our guide to SEO backlink analysis and strengthening a website's authority.
What you will need
access to a Search Console property, if the site belongs to you;
a third-party index tool, essential for a competitor and useful as a complement for your own site;
a spreadsheet to consolidate the exports, with one sheet per source;
a written objective, and a folder in which to keep each dated export.
Step 1: collect the backlink data
What to do: for each available source, export the list of links and domains pointing to the site, noting the collection date. The details vary depending on whether or not the site is your own.
The Search Console Links report
Open the Links report for your property. It provides a link profile report: the sites that link to your site most, the most-linked pages and samples of external links. It is only available for a property to which you have access.
It is also the most direct answer for anyone looking for a free analysis of their own site: Search Console costs nothing. For a site that is not yours, it is of no use.
Its limitations
This report is not an exhaustive list. According to Google's help documentation, it displays a sample of the links found, with groupings, and its tables are limited to 1,000 rows. Some URLs may be omitted, notably if they are not indexed or if they have been deduplicated. Target pages, meanwhile, are grouped by canonical URL, that is, the reference address chosen when several addresses show the same content.
The consequence is simple to state and easy to forget: a drop in links, or a link's absence from the report, does not on its own prove that the link has disappeared from the web. It may stem from sampling, grouping or different indexing.
Analysing a competitor with a third-party index
For a competitor, use a third-party index tool. It can display backlinks, referring domains, anchors, attributes, and the new or lost links it has detected. Remember the verb: it has detected. The tool observes the links it has itself discovered, on the dates it discovered them.
Its data therefore depends on the size of its index and its crawl rate. It never replaces what Search Console gives a site's owner. On your own site, cross-check the two sources: a link seen by one and not the other deserves a closer look, not a hasty ruling.
Good to know
Name each export with its source and collection date. In a few months, this is what will let you tell a genuine change from a mere change of snapshot.
Step 2: map the link profile
What to do: merge your exports into a spreadsheet and structure the columns before drawing any conclusions. Keep the observations specific to each link: a single source page may contain several links, and a single referring domain may link to several pages on your site. A clean map stops you relying on totals that flatter or alarm.
When several sources report the same backlink, match their observations using the source URL, the target URL and, where available, the anchor or the link's position. Keep the provenance and dates of each source, without counting the same observation twice. Then build two views: the breakdown by backlink to examine anchors and pages, and the summary by referring domain to measure diversity and concentrations.
Backlinks and referring domains
A backlink is a link to the analysed site. A referring domain is a domain that sends at least one backlink. A single domain can therefore send several backlinks while counting as only one referring domain.
Analyse the two counts separately. A large number of backlinks coming from few domains tells a story (a repeated footer link, for example) very different from the same quantity spread across many distinct domains.
Target pages and anchors
Structure the table with the following columns: source URL, target URL, anchor (the clickable text of the link), first observed, last observed and status. Each link is thus tied to its page of origin, the page it points to and the period during which it has existed.
Then sort by target URL to see which pages attract links, and by anchor to spot those that are over-represented. A heavily linked page is not a problem in itself; it is often useful content. An anchor repeated identically across many links, however, calls for a manual check.
New and lost links
Compare the first and last observed dates to isolate recent links and those that have disappeared. Favour long-term trends over isolated movements, and check important links by hand before concluding that one has been lost. Remember that a link's absence from an export may be due to the source's limitations.
The sponsored, ugc and nofollow attributes
Add a column for the link's rel attribute. Google recommends sponsored for advertising or paid links, and ugc for user-generated content. nofollow is used when no other attribute fits, or when the site does not wish to vouch for the linked page.
These attributes indicate the declared nature of the link. They say nothing about its value or its danger: a link with no attribute may be worthless, and a nofollow link may be perfectly legitimate.
Step 3: assess the quality of the backlinks
What to do: select the most influential and the most suspicious links, then examine them one by one in context. Volume is no substitute for quality, and a spreadsheet does not replace reading the source page.
Relevance and editorial context
Google relies on links, their anchor text and their context to understand the relevance of pages, and recommends links that make sense to users and to the search engine alike. For each link you select, open the source page and ask yourself why the author created that link. A clear editorial reason is the best sign of a healthy link.
Rather than a single score, apply a simple framework:
Criterion | Question to ask | Signal that calls for a check |
|---|---|---|
Topical relevance | Does the topic of the source page match that of the linked page? | No apparent connection between the two pieces of content |
Editorial context | Does the link serve the reader within the text? | An isolated, tacked-on link in the middle of a list of links unrelated to the subject |
Natural anchor | Does the anchor describe the target page as an author would? | Identical wording repeated across many pages |
Consistency with the target page | Does the linked page deliver what the anchor promises? | Anchor and content that do not match |
Authority scores: filters, not verdicts
The authority scores offered by tools are proprietary metrics: the vendor calculates them using its own data. They are not an official Google metric, and must never be presented as a rating that Google would assign to a site.
Their reasonable use is as a sorting filter: ranking links to decide which to review first. They must then be supplemented by examining the source site and the actual link. A high score does not guarantee a legitimate link, and a low score does not make a link harmful.
Step 4: spot risky links and decide what to do
What to do: check suspicious links against Google's rules, then choose the least risky action. In most cases, the right decision is to do nothing drastic.
Link schemes that breach the guidelines
For Google, link spam refers to links created primarily to manipulate rankings. This includes buying or selling links that pass ranking benefits, excessive link exchanges, automated links, and certain links distributed at scale through templates, widgets, directories or comments.
One important nuance protects many sites: advertising or sponsored links do not breach the policy if they are properly qualified, for example with rel="sponsored" or rel="nofollow". The problem is a link that claims to be editorial when it is not.
When to disavow, and when to hold back
A disavow file is a file you submit to Google asking it to ignore certain links. According to Google, it should only be considered when a large volume of spammy, artificial or low-quality links has caused, or is likely to cause, a manual action. Google also states that most sites do not need one.
First, try to have the links in question removed. Disavowing does not remove anything from the web: it simply flags links that should not be taken into account.
Point of caution
A poorly prepared disavow file can harm a site's visibility. Treat it as an advanced measure, to be used with care and never as routine housekeeping.
If a manual action is reported
If the Manual actions report in Search Console flags artificial links, review the links concerned, document where they come from, and first try to have them removed or properly qualified. If some removals are impossible, a disavow can form part of the remediation process, before a possible reconsideration request. Keep a record of the steps you have taken: it will help justify your decisions. The full procedure calls for a dedicated diagnosis of toxic backlinks.
Traffic drop after a spam update
A traffic drop after a spam update does not, on its own, justify a mass clean-up. Google notes that neutralising spammy links does not necessarily restore the ranking benefits they provided. Examine the context first before attributing the decline to backlinks.
Worked example: auditing the link profile of a fictional blog
Here is an illustrative, entirely fictional scenario: an editorial blog of a few hundred pages, which we will call “Carnets de Marée”. Its manager has access to Search Console and an export from a third-party index tool. No performance figures are given, deliberately.
Consolidation. He creates a spreadsheet with one dated sheet per source. He then merges the observations into a single table with the columns source URL, target URL, anchor, attribute, first and last observation, status and provenance. Two exports may report the same link: he groups them without losing the dates and sources. However, he keeps distinct backlinks separate, even when they come from the same domain. A second view, aggregated by referring domain, will make it possible to measure link concentration without overwriting the information for each source page.
Backlinks and referring domains. Counting separately, he finds that a partner site accounts for a large share of the total backlinks, through a sidebar link repeated on each of its pages. Yet this partner counts as only one referring domain: the total number of backlinks gave a misleading picture of the profile's diversity.
Heavily linked page and over-represented anchor. Sorting by target URL, he sees that an old guide attracts a large share of the links: it is content that is cited spontaneously, nothing worrying. Sorting by anchor, an exact-match phrase recurs across several pages of the same domain. He flags it for manual checking.
Manual review. He selects ten priority links, chosen by sorting on authority score, but also because they are suspicious or recent. For each one, he opens the source page and applies the grid: topical relevance, editorial context, natural anchor, consistency with the target page.
Decision. Most links are editorial citations: no action needed. For the partner's sidebar links, he requests appropriate qualification if the relationship is commercial. For a link from an automated comment, he attempts a removal request. Disavowing remains a last resort, set aside here as long as no manual action is confirmed or likely.
Worth noting
A cautious conclusion might read: “Among the ten links examined, nothing establishes the existence of a widespread manipulative pattern. No disavow decision is made at this stage. The doubtful links will be subject to further checks, and the Manual actions report in Search Console will be consulted separately to see whether Google has flagged a violation.” The sample guides the audit; it cannot on its own be used to judge the whole profile.
Common mistakes to avoid
Treating the Search Console report as an exhaustive list. This is a sample: a simple drop does not prove a loss of links.
Confusing backlinks with referring domains. A backlink total inflated by a single domain says nothing about the diversity of the profile.
Treating a tool score as a Google criterion, or judging a link on its attribute alone. Neither tells you anything about a link's real value or danger.
Disavowing in bulk as a reflex, without a confirmed or likely manual action and without a prior attempt at removal.
Practical tips for a lasting audit
A backlink audit is not a one-off task. Set a regular review schedule, adapted to how quickly your site gains or loses links, and track new and lost links each time. Keep all your exports dated: they allow you to compare trends rather than snapshots.
Document your decisions as you go: links contacted, replies received, qualifications requested. This record will save you valuable time if a reconsideration ever becomes necessary.
Finally, do not confuse backlink analysis with website content analysis. A link audit is only one part of a broader analysis, which also looks at content, technical aspects and visitor experience. A healthy link profile sheds light on one part of the picture; it does not excuse you from looking at the rest.
Also read: Backlink: definition and role in SEO.
FAQ
For your own site, yes: the Search Console Links report is free and restricted to properties you have access to. For a competitor's site, you need a third-party index tool, whose free features vary by provider.