Backlink audit: reading your link profile without panic

Backlink audits have a reputation for producing anxiety. A tool scans your profile, colours a third of your links red, labels them "toxic" and recommends disavowing them. In most cases that advice is unnecessary, and occasionally it is harmful. Every site with any visibility accumulates links from scrapers, spam directories, auto-generated pages and strange foreign domains. Google has said its systems are designed to ignore that kind of link.

A sensible audit has three goals: understand where your genuine authority comes from, identify any patterns that could really be a problem (almost always the result of past link buying or schemes), and find opportunities to grow. This guide covers each.

Gather the data

• Search Console Links report: top linking sites, top linked pages and top anchor texts, with an export. It is a sample, but it is Google's own view.
• One commercial backlink tool: more complete lists, history, and first-seen and lost dates. Different tools find different links; one is usually enough.
• Your own knowledge: past agencies, campaigns, sponsorships, guest posts and any link buying. The audit is far easier if you know what was done in the site's name.

Work at the level of referring domains, not individual links. A sitewide footer link from one partner can show up as 30,000 links; it is one relationship.

What a normal profile looks like

Natural link profiles are messy. Expect to see:

• A minority of strong, relevant domains that account for most of the real value.
• A long tail of small blogs, forums, directories and social pages.
• A noticeable share of junk: scrapers that copy your content, spam directories, auto-generated "website worth" and "SEO analysis" pages, foreign-language spam.
• Anchor text dominated by your brand name, your URL, and generic phrases ("here", "this guide", "website").
• Links pointing mostly at the home page, with the rest spread over content that people find useful.

The junk is not a problem in itself. It accumulates for everyone, and Google is well practised at ignoring it.

Patterns that do deserve attention

Real risk comes from links that look like someone was trying to manipulate rankings. The tell-tale patterns:

Pattern	What it suggests
Many links with exact-match commercial anchors ("cheap hiking boots") from unrelated sites	Bought links or a link scheme
Clusters of sites with similar designs, the same hosting, thin content, all linking to you	A private blog network
Guest posts on "write for us" sites covering every topic, each with a keyword link to you	Guest posting at scale for links
A sudden surge of links in a short period with no PR or viral reason	A campaign someone ran, or a negative SEO attempt
Links from pages that are clearly link-selling (e.g., "sponsored" lists of unrelated links)	Paid placements without rel attributes
Sitewide links in footers of unrelated sites with keyword anchors	Widget, template or paid sitewide links

If you find these and they were built by you or someone acting for you, that is a real risk, especially in competitive niches. If you find a few of them and you did not build them, they are most likely ignored.

Tool metrics, and what they do not mean

Commercial tools give each domain a score for authority or rating, and some give each link a spam or toxicity score. These are useful for sorting long lists. They are not Google's numbers and Google does not use them. A domain with a modest score can be the most relevant link you have, such as a niche association site, and a domain with a high score can be irrelevant to your business. Use the scores to decide where to look first, then judge each important domain by visiting it: is it a real site with real readers, and does it make sense for it to link to you?

A worked example

A hypothetical home-insurance comparison site runs an audit after a gradual decline. The tool flags 1,900 of 3,100 referring domains as toxic. Looking closer, about 1,700 of those are scrapers, spam directories and auto-generated SEO pages that every site in the niche has; they are ignored. The remaining 200 are different: guest posts on general "lifestyle" blogs, all published within eighteen months, nearly all with anchors like "cheap home insurance quotes". The marketing team confirms a previous agency bought them.

That pattern is a genuine risk. The team asks the blogs to remove or nofollow the links (a minority comply), disavows the rest at the domain level, and records the work. The 1,700 junk domains are left alone. Separately, the audit shows that the site's best links come from two consumer-advice charities and a regional newspaper, all pointing at a single guide about claims; that guide becomes the template for the next content projects.

Reading the anchor text distribution

Export anchor texts by referring domain and group them into categories: brand (including variations), naked URL, generic ("click here", "website"), topical (describing the subject without being a target keyword), and exact-match commercial keywords. There is no correct percentage, but a profile where commercial exact-match anchors make up a large share across unrelated domains is the most common sign of past manipulation. The anchor text and disavow guide covers what to do about it.

Finding where your real authority comes from

The more useful half of an audit is positive. Sort referring domains by relevance and quality (your own judgement plus tool metrics as a hint) and list your best twenty. For each, note which of your pages it links to and why. This tells you:

• Which content earns links naturally, and therefore what to create more of.
• Which relationships (partners, media, associations) produce links, and therefore where to invest.
• Which of your pages hold the most link value, and should link internally to the pages you want to rank. See internal linking.

Finding opportunities

• Links to dead pages: targets returning 404 or redirecting badly. Fix them first; the reclaiming links guide explains how.
• Lost links: links that disappeared recently from good sites. Sometimes a page was updated and the link removed by mistake; a polite note can restore it.
• Competitor gaps: relevant sites linking to two or three competitors but not to you. These are often directories, associations, resource pages and media that cover your industry.

When action is warranted

Most audits end with no disavow at all. Action makes sense in three situations:

1. You have a manual action for unnatural links in Search Console. Then removal and disavow are required; see manual actions.
2. You know links were bought or built through schemes in the site's name, at a scale that could trigger action, and you want to clean up proactively.
3. You see a clear manipulative pattern that you believe is being counted against you, and there is no other explanation for persistent problems.

In those cases, try removal first (contact site owners), then disavow at the domain level for what remains. Do not disavow links simply because a tool scored them badly.

A half-day audit routine

1. Export Search Console links and one tool's referring domains; merge on domain.
2. List the top twenty genuine referring domains and the pages they link to.
3. Group anchor texts by category; look for exact-match clusters.
4. Skim the remaining domains quickly for manipulative patterns, ignoring obvious junk.
5. Check target URLs for 404s and broken redirects.
6. Compare with two competitors to find relevant gaps.
7. Write a one-page summary: strengths, risks (if any), fixes, opportunities.

Keep the summary and the merged export. Next time, compare against them: new strong domains, lost ones, and any new pattern stand out immediately. That summary is more valuable than any toxicity score. It tells you what to protect, what to clean up, and where your next good links are likely to come from.