Link extractor: list every link on a page

How to use the tool

Paste the full address of a page and click “Extract”. The tool downloads the page the way a browser would, follows redirects to the destination, then picks up every <a> tag in the HTML it receives. If the page is protected, or if you want to test code before publishing it, open “Paste HTML instead”: the analysis then happens entirely in your browser and nothing is sent.

The result can be filtered by type, by the presence of a rel attribute, by empty anchor or by link that is not crawlable, and sorted by anchor or by address. The CSV file contains the rows on display, with the address as written in the code and its full address. The table shows only the first 500 rows of a filtered list; the CSV holds all of them.

What each row contains

• Anchor. The text of the link. When a link wraps an image with no text, the tool takes the image’s alt attribute, as Google does, and says so under the anchor.¹ Failing that, it keeps the title attribute, which Google may use when text is missing. A link with no text, no alt and no title is marked “empty”.
• Address. The full address, computed from the original page or its <base> tag. When the code holds a relative address, the address as written appears below.
• Type. Internal when the host name is the page’s own, “www.” aside; external otherwise, subdomains included. Links that target a fragment of the same page are classed as “same-page anchor”. mailto: and tel: links fall under “other schemes”.
• rel. The attribute values, separated: nofollow, sponsored, ugc, noopener, noreferrer.
• target. Where the link opens. _self is the default; _blank usually opens a new tab.³
• Area. The nearest structural tag around the link, among nav, header, footer, aside and main. “Other” when none surrounds it. A header placed inside an article counts as a header: the area is deduced from tags and ignores the visual layout.

What Google expects from a link

Google reliably follows only an <a> element with an href attribute that resolves to a web address.¹ The same document advises against javascript: links, framework attributes such as routerLink and elements made clickable by onclick without an href: Google cannot reliably extract addresses from them. The tool therefore marks as “not crawlable” a link with no href, an empty href, an href reduced to #, or a javascript: one. MDN agrees on the usage side: a fake href such as # or javascript:void(0) breaks opening in a new tab, bookmarking and screen readers, and a button suits an action that leads to no address.³

Anchor text matters too. Google asks for descriptive, concise text that can be read on its own, out of context, and advises against empty or generic text such as “Click here”. For a linked image, the alt attribute serves as anchor text, and an empty alt is to be avoided.¹ The tool does not judge the quality of a text: it flags the absence of text, which you can isolate with the “Empty anchors” filter.

On the number of links, Google’s documentation is measured: every important page should contain a link to at least one other page of the site, there is no ideal number, and an excess is probably a problem.¹ The count summary above the table gives the total, the distinct addresses and the split between internal and external, enough to situate a page.

nofollow, sponsored, ugc: reading the rel column

An ordinary link needs no rel attribute. The three values serve the other cases.² sponsored qualifies paid links, ads or purchased placements. ugc qualifies links that come from user-generated content, comments and forum posts. nofollow remains the value to use when neither of the others fits and you do not want Google to associate your site with the linked page or to crawl it from your site. Values can be combined, for example rel="ugc nofollow".²

Google presents these attributes as hints: links that carry them are generally not followed, and the linked page can still be discovered through a sitemap or links from elsewhere.² For a link to a page on your own site, Google recommends the disallow rule of robots.txt instead.²

The noopener and noreferrer values belong to another subject, the safety of opening in a new tab. According to MDN, target="_blank" implicitly gives the behaviour of rel="noopener", and noreferrer adds that the Referer header is not sent.³⁴ These values appear in the column but carry no weight in search.

Checking the HTTP status of ten links

Tick up to ten internal or external links in the table and click “Check the selected links”. The tool queries the ten addresses in parallel, in a single request from your quota, and shows the status code of each first response. It does not follow redirects: a 301 appears as 301, with its destination, which helps spot links to update. The body of the response is not read.

An address over 300 characters cannot be ticked, and a site that does not answer in time is marked “no answer”. To follow a redirect chain to its end, the redirect and HTTP status checker takes over.

What the tool does not see

The tool reads the HTML as the server sends it, without running JavaScript. Links that a script adds after loading, in a dropdown menu or a list loaded on demand, are therefore missing, and a page built entirely in the browser returns few links, or none. Links inside modal windows or tabs are present if they are in the original code.

Forms, XML sitemaps, links from <link> tags (canonical, hreflang, feeds) and images with no link are not <a> tags: they fall outside the scope. So do the <area> tags of image maps.

After the extraction

A page that receives few internal links is hard to find for Google and for visitors alike. The internal linking guide explains how to spread those links, with anchors that say where they lead. To inspect the headers returned by an address found in the list, the HTTP headers analyser gives all those of the response. And when the result shows hundreds of links to rework across a whole site, it is time for a full SEO audit, which prioritises the fixes.

Frequently asked questions