The Search Console Links Report: What It Hides

The Search Console links report is a free list of the pages on your site that other sites link to, the sites doing the linking, and the pages on your own site that link to each other. It lives under Links in the left-hand menu, and it is the only link data that comes straight from Google rather than from a third-party crawler. That is its real value, and also the source of its main limitation: it is a summary Google chooses to show you, not a complete index of the web graph.

This post is about reading that report properly. It covers what each of its tables contains, what is missing from them, how they behave when you export them, and how to fold them into an audit without over-trusting any single number. It does not try to be a link-building guide or a tour of the whole of Search Console.

What the Links Report Is Made Of

The report has two halves, External links and Internal links. Each half has its own tables, and the two answer different questions, so it helps to keep them apart in your head.

Under External links you get three tables. Top linked pages lists the URLs on your property that receive the most links from other sites. Top linking sites lists the domains that link to you most often. Top linking text lists the anchor text used in those links. Under Internal links there is one table, Top linked pages, which ranks your own URLs by how many internal links point at them from elsewhere on your site.

Each table is sorted by link count by default. Click any row and you drill down: from a linked page you can see which sites link to it, and from a linking site you can see which of its pages link to you and which of your pages they point at. That drill-down is where most of the practical value sits, because the headline tables alone tell you very little about why a link exists.

The report is scoped to the property you have selected. A domain property aggregates across subdomains and protocols, while a URL-prefix property only reports on that prefix. If you see a surprisingly small list, check which property you are in before assuming anything about your link profile.

Reading Top Linked Pages Without Fooling Yourself

The external Top linked pages table is a map of where your earned attention actually lands. In most sites it is lopsided: the home page takes a large share, a handful of articles or tools take most of the rest, and the commercial pages that you most want to rank have few or none. That mismatch is the single most useful thing the table shows you.

Count is not quality. A page with four hundred links from one scraper network and a page with six links from relevant, independent publications will show up in the wrong order. Use the table to find candidates, then open the drill-down to see the distribution: many links from a few sites, or a few links from many sites. The second pattern is usually healthier and more durable.

Watch for URLs that should not be getting links. If a parameterised URL, a staging path, an old campaign page or a redirect source appears near the top, people are linking to an address that is not your preferred one. Check whether it redirects or carries a canonical tag pointing to the right page. A link to a URL that 301 redirects generally still helps, but a link to a URL that returns an error helps nobody, and the table is how you find those.

A practical routine is to export the table and compare it against your list of pages that matter commercially. Anything that has earned links but returns a 404, a soft 404 or a thin page is a recovery opportunity. Restore the content, or redirect the old URL to the closest equivalent page, and the link value has somewhere to go again.

• Sort out the pages with links but no rankings first: they show where authority exists and is being wasted.
• Flag any linked URL that does not return a clean 200 or a deliberate redirect.
• Note which linked pages are old posts. They often make the best places to add an internal link to a page you want to lift.

Top Linking Sites and Anchor Text

Top linking sites is grouped by domain, not by page, which is why a single site with a sitewide footer or sidebar link can dominate it. The count there means links, and one domain can contribute thousands of them while representing a single editorial decision. Do not read a large number as a large endorsement.

What the table does well is let you spot patterns quickly. A long tail of unfamiliar domains with very high link counts is worth a closer look in the drill-down: check whether the linking pages look like real content, directory dumps, scraped copies or auto-generated pages. A cluster of links from a partner or a former agency is also common and usually explains an otherwise puzzling anchor distribution.

Top linking text shows the most common anchors. For a normal site you expect the brand name, the bare domain, the page title and some generic phrases such as click here or this article. A list dominated by exact-match commercial phrases is unusual for a naturally earned profile and deserves a manual review, although it is not proof of anything by itself. Bought or traded links tend to leave anchors that look that way, and search engines are explicit that manipulative link schemes violate their guidelines.

The report does not tell you whether a link is followed, nofollow, sponsored or user-generated, and it does not show you when a link was first discovered. You cannot read a clean attribute breakdown from it. If you need that, you have to visit the linking page or use a crawler that records link attributes.

What the Report Hides

The limits matter more than the features, because they decide what conclusions you can safely draw.

First, it is not exhaustive. Google presents this as a sample of the links it knows about, and the numbers will rarely match a third-party backlink tool. Different crawlers discover different pages, keep links for different lengths of time and define a linking page differently, so two tools can disagree by a wide margin without either being wrong. Treat the Search Console figure as one view, not as the truth.

Second, the tables are capped. You see the top rows, not every linking site, so a long tail is cut off. That means you cannot use the report to prove that a link does not exist. Absence from the list is not evidence of absence from Google's index of links.

Third, there is no history. You get a current snapshot with no graph of links gained and lost, no first-seen date and no way to see which links vanished last month. If you want trends, export on a schedule and keep the files yourself. A monthly export into a spreadsheet costs almost nothing and gives you a record that the interface never will.

Fourth, it does not reflect your disavow file. Links that you have disavowed generally still appear in the report. Their presence does not mean the disavow failed. Disavowing is also a last resort for clear link-scheme problems, not something to do because a domain looks ugly.

Fifth, it tells you nothing about value. There is no authority score, no traffic estimate for the linking page and no topical relevance rating. Any judgement of quality is yours to make by looking at the linking page.

The Internal Links Table and What It Reveals

The internal half is often more actionable than the external one, because every problem it reveals is something you can fix yourself today. The Top linked pages table under Internal links ranks your URLs by the number of links from other pages on the same property.

Read it with the structure of your site in mind. Your home page, main category pages and navigation targets should sit near the top. If a thin tag archive, a login page or a privacy policy outranks your key service pages, your linking is telling Google that those pages are the centre of the site. Sitewide links in a footer inflate counts for exactly those utility pages, which is normal, but it is worth knowing.

Now look at the bottom and the missing. The table is sorted with the most-linked pages first, so scrolling to the end or exporting it shows the pages with the fewest internal links. A page that matters to you but has one or two links, all from low-importance pages, is an orphan in practice even if it technically has an inbound link. Cross-check against your sitemap: any URL in the sitemap that does not appear in the internal links table at all is a true orphan.

A worked example. You run a site with forty articles and a /services/ section. The internal table shows the blog index with 300 links, the contact page with 290 and the flagship service page with 6, all of them from old blog posts. The fix is small: add contextual links from the ten most relevant, best-performing articles to that service page, using descriptive anchors, and link to it from the services overview. You have changed the internal signal without touching anything external.

Keep in mind that this table counts links discovered in the pages Google crawled, so JavaScript-injected links that render late, or links in areas Google does not crawl, may be missing or undercounted. If a link you know exists is not reflected, test the rendered HTML with the URL Inspection tool.

Using the Report in an Audit

A reasonable audit using only this report takes under an hour and follows a fixed order.

Start with the external Top linked pages export. Check the response status of each URL, and mark anything that is a 404, a redirect chain or a parameter variant. Fix the dead ones first, because reviving a linked page is the cheapest link-building you will ever do.

Next, look at Top linking sites. Sort out the recognisable, relevant publishers, the obvious noise and the unknowns. Open a sample of each group. Write down anything that looks like a paid or exchanged placement so you can decide how to handle it, rather than reacting to a number. If you have received a manual action notice for unnatural links, that is a different process, covered in the Manual actions report, and it calls for evidence, not a guess based on this table.

Then scan Top linking text for anything that is out of balance: a single commercial phrase repeated heavily, or anchors in an unrelated language or topic. Those are signs of either a hacked-page injection on someone else's site or a past campaign that got out of hand.

Finish with the internal table. Compare it against your priority pages, your sitemap and your navigation, and list the pages that need more or better internal links. Combine this with the Performance report: a page that has impressions but sits outside the top results, and has very few internal links, is usually the quickest win available.

Record the date and the numbers at the end. Because the interface has no history, the export you save today is the baseline you will compare against in three months.

• Export all four tables and date the files.
• Check status codes for every externally linked URL.
• Review anchor text and the drill-down for unusual linking sites.
• Compare internal link counts to your priority pages and sitemap.
• Re-export on a fixed schedule and diff the results.

Common Mistakes When Reading It

The most common error is comparing Search Console totals to a paid tool and concluding that one of them is broken. They measure different samples. Choose one source for trend-watching and stick with it; use the other for cross-checking specific links.

A second is chasing link counts as a goal. A rising number of links from the same two sites is not progress. What matters is the number of distinct, relevant sites and where the links land.

A third is disavowing on sight. An unfamiliar domain is not automatically a risk. Search engines are generally good at ignoring low-quality links that nobody paid for, and acting against a clean profile can remove links that were helping.

A fourth is ignoring the internal table because it feels less glamorous. It is the part of the report where the effort-to-result ratio is best, since you control both ends of every link.

Finally, remember that access is tied to verification. Users with only restricted access may see less, and a property that was verified recently will not show history from before. If data looks absent, confirm permissions and the property type before drawing conclusions.