Technical
Anchor Text Extractor
See the text people actually click
Give it a page URL and get every anchor on it grouped by destination, with the counts, the generic phrases separated from the descriptive ones, the empty and image-only anchors listed, and the sentence each link sits in.
Get early accessWhat it returns
- Every anchor string on the page with a count, grouped under the URL it points to, so one destination described five ways appears as five variants of the same link.
- Generic anchors pulled into their own list: click here, read more, this page, learn more, and the bare URL used as its own label.
- Anchors with nothing to read: image links reported with their alt attribute, icon links reported as empty when no text or label exists at all.
- The sentence each anchor sits in, so you can judge whether the text makes sense to a reader without opening the page and hunting for it.
When this matters
Somebody audits a help centre and finds the same setup guide linked from forty articles, thirty of them with the anchor 'here'. Nothing is broken and nothing will ever be reported, but a reader skimming for the link gets no signal about where it goes, and neither does anything else parsing the page. Pulling the anchors is also the fastest way to catch the opposite mistake, the same exact keyword phrase used as the anchor on every internal link, which reads as written for a machine because it was.
How to answer this today
The anchor text that moves most is usually not your own. Backlinks reports the text other sites use when they link to you, grouped by phrase and by linking domain, which is the view for judging how the rest of the web describes a page. Reading the anchors on one page you control, on demand and with no project, is the part this tool adds.
BacklinksWhy this is harder than it looks
Anchor text is not only the text
A link can carry an image with alt text, a nested span, a visually hidden label, or an aria-label that a screen reader announces instead of the visible words. Reading the text node alone reports an empty anchor for a link that is actually described perfectly well.
Whitespace and case invent variants
'Read More', 'read more' and 'read more' are one anchor to a reader and three rows in a naive grouping. Normalising too hard hides a real difference between two anchors, normalising too little manufactures one, and the choice has to be visible in the output.
Template links dominate the counts
A header, a footer and a sidebar put the same anchors on every page of a site, so an unfiltered count mostly describes the template rather than the article anyone wanted to read. Separating body anchors from boilerplate is what makes the list worth looking at.
Rate limits
Reading the anchors means fetching the HTML of a page we do not own, one request at a time. A per-visitor limit stops the tool being driven as a bulk scraper against a single site, and keeps the fetcher off the blocklists that would make it useless for everyone.