Anchor text is the visible, clickable text of a hyperlink, and search engines read it as a description of the page the link points to. This article defines anchor text, explains anchor text as a relevance signal, and sets out exact n-gram reuse, in which the anchor repeats the target page’s head phrase. It then covers anchor variety without dilution, anchor text in the bridge and supplementary sections of an article, and anchor text on holisticradar.com as a worked example. It closes with anchor text in semantic SEO services, followed by image link alt text, navigation anchors, and over-optimization.
What anchor text is
Anchor text is the visible, clickable text between the opening and closing tags of an HTML link, such as the words “Technical SEO services” in a link whose href attribute points to /services/technical-seo/.
A link carries four pieces of information, and anchor text is only one of them. The href attribute names the target. The anchor text labels the target. The words around the link supply its context. The rel attribute, where present, qualifies the relationship, for example marking a link as sponsored or as user generated. A search engine reads all four, but the anchor text is the part written specifically to say what the target is.
Anchor text is classified by how closely it matches the name of the target page:
| Anchor type | Example for a page titled “Canonical Tags” | What the label tells a search engine |
|---|---|---|
| Exact match | canonical tags | The target is about exactly this term |
| Partial match | how canonical tags handle duplicates | The target is about this term, with an added aspect |
| Branded | Holistic Radar | The target belongs to this entity, not which subject it covers |
| Naked URL | holisticradar.com/blog/canonical-tags/ | Only what the URL slug itself says |
| Generic | click here, read more | Nothing about the target |
| Image | A linked image with alt text | Whatever the alt attribute says |
The anchor text is a property of the linking page, not of the target. The target page cannot set the words other pages use to point at it; each linking page chooses its own label. On a site’s own internal links, the site writes every label, which makes anchor text one of the few relevance signals a site controls completely.
Anchor text as a relevance signal
Anchor text is a relevance signal because it describes the target page in words chosen by the page that links to it, rather than by the target page itself.
The idea is as old as Google. The 1998 paper in which Sergey Brin and Larry Page described the Google prototype explained that the system associated a link’s text with the page the link pointed to, because anchors often describe a page more accurately than the page describes itself, and because they let the engine index pages it had not yet crawled. A page that many links call “canonical tags” is, by that evidence, a page about canonical tags.
Google’s current guidance on links states the same principle in practical terms. Good anchor text is descriptive, reasonably concise, and relevant to both the page it sits on and the page it links to. The guidance warns against generic anchors such as “click here”, against stuffing keywords into anchors, and against chaining several links together so that none of them has words of its own around it. It also notes that the words before and after a link matter, which means the sentence around an anchor is read as part of the label.
Anchor text serves people in the same way. A screen reader can list every link on a page out of context, and the Web Content Accessibility Guidelines require that a link’s purpose be clear from its text, or from its text together with its immediate context. An anchor that makes sense when read alone serves a search engine, a screen reader user, and a scanning reader at once.
Exact n-gram reuse
Exact n-gram reuse is the practice of writing anchor text that repeats the target page’s head phrase word for word, so that the label on the link and the name of the page are the same string.
An n-gram is a contiguous sequence of n words: “canonical tags” is a two-word n-gram, and “technical SEO services” is a three-word n-gram. The head phrase of a page is the n-gram that names its subject, which on a well-built page is the leading term of the title and the H1. Exact n-gram reuse takes that phrase and uses it, unchanged, as the anchor.
| Target page | Head phrase | Exact reuse anchor | Drifting anchor |
|---|---|---|---|
| /blog/canonical-tags/ | Canonical tags | canonical tags | fixing duplicate URLs |
| /services/technical-seo/ | Technical SEO services | Technical SEO services | our site health work |
| /services/conversion-copywriting/semantic-copy/ | Fact-first writing | Fact-first writing | a clearer way to write |
The drifting anchors are not wrong descriptions. Each one points toward the right subject. The problem is that each one introduces a second name for the page, and a page with several names has a weaker claim to any of them. Exact reuse gives every link to a page the same label, which confirms the page’s subject with each link instead of spreading it across synonyms.
Exact reuse follows four rules:
- Take the phrase from the target. The anchor is copied from the target page’s title or H1, not composed by the writer of the linking page.
- Keep the term form. Singular stays singular and plural stays plural; “canonical tag” and “canonical tags” are different n-grams.
- Add nothing inside the phrase. Qualifiers such as “best”, “complete”, or “ultimate” break the n-gram and add a claim the target page does not make.
- Change the target, then change the anchors. When a page is renamed, every anchor pointing to it is updated to the new head phrase.
Anchor variety without dilution
Anchor variety without dilution means that every anchor pointing to a page contains that page’s head phrase, while the words around each anchor change with the context of the linking page.
Variety and consistency apply to different parts of the link. The anchor itself stays constant, because it names the target. The sentence around it varies, because each linking page has its own reason to send the reader on. A page about featured snippets and a page about structured data can both link to the same core page with the same anchor, and each surrounds it with a different justification. The target receives one name and many contexts, which is the combination that tells a search engine both what the page is and which neighbouring subjects it relates to.
Dilution happens in two patterns:
| Pattern | Example | Effect |
|---|---|---|
| One target, many labels | Links to the technical SEO page read “site audits”, “speed work”, and “technical help” | The page’s subject is spread across several names, none of them dominant |
| One label, many targets | The anchor “SEO audit” points to a service page on one page and to an article on another | The label no longer identifies a single page, and the two targets share one name |
External links behave differently. A site does not write the anchors other sites use, and a natural backlink profile contains branded, naked, generic, and partial anchors in a mix the site does not control. Variety in external anchors is a sign of independent linking; variety in internal anchors is a sign that the site has not decided what its own pages are called.
Anchor text in the bridge and supplementary sections
Holistic Radar places every contextual anchor in an article’s bridge section and supplementary section, and the main content carries no links, so the main sections stay on the article’s one contextual vector.
The reason follows from how anchors are read. A search engine interprets an anchor together with the words around it, and in the main content those words describe the current article’s own subject. A link placed mid-definition borrows that context for a different target and invites the reader and the crawler to leave before the answer is complete. Removing links from the main content keeps every main section about one subject and keeps every reader on the page until the subject is covered.
The bridge and supplementary sections are written for the opposite job. Their surrounding text is about the relationship between this article and another page, so an anchor placed there sits inside words that describe its target.
| Section | Links | Anchor rule |
|---|---|---|
| Extractive summary | None | Not applicable |
| Main sections | None | Not applicable |
| Worked example | None | Not applicable |
| Bridge section | Exactly one, to one core page | Anchor equals the core page’s name |
| Supplementary section and next reads | At most three, to other articles | Anchor equals the target article’s title, in its head-phrase form |
Each anchor in this structure is paired with a one-line reason that states why the reader should follow it. The reason supplies the context that a generic sentence would otherwise have to carry, and it keeps the anchor itself free to be nothing but the target’s name.
Anchor text on holisticradar.com
On holisticradar.com, every bridge anchor equals the name of the core page it points to, and every next-read anchor equals the title of the target article in its head-phrase form, the words before the colon.
The next reads at the end of this article show the rule: the article titled with the head phrase “Internal linking” is linked with the anchor “Internal linking”, and the same holds for “Semantic content network” and “Topical map”. The promise that follows the colon in a title is left out of the anchor, because it describes what the article delivers rather than what the article is called.
The bridge from an article about featured snippets uses the anchor “Fact-first writing”, which is the name of the service node it links to. The bridge from an article about canonical tags or structured data uses “Technical SEO services”. The bridge from this article uses “Semantic SEO services”. No bridge anchor describes the core page in other words, so every article that bridges to a core page repeats the same label for it.
The same discipline applies across the site’s five internal link types: hub to node, node to hub, lateral, contextual bridge, and forward. Navigation and breadcrumbs sit outside that set, and the editorial links, the contextual bridge and the supplementary next reads, always use the target’s own name as the label. Combined with the rules that every article has a single bridge link, at most three supplementary links, and zero outbound links, the result is a link set in which every anchor can be predicted from its target alone.
The rule also simplifies maintenance. When a page’s name changes, every anchor that points to it is found by searching for the old name, because no anchor ever used a variant of it.
Anchor text in semantic SEO services
Anchor text is the point where a site’s naming decisions become visible to a search engine. The label on each link can only be consistent if every page has already been given one head phrase, and that phrase can only be chosen well if the query network shows which term form searchers use and the topical map shows which page owns it.
Semantic SEO services from Holistic Radar make those decisions first. The topical map assigns each attribute to one URL and names it, and the internal link plan then fixes every anchor to the name of its target. Anchor text written without that groundwork has nothing stable to repeat.
→ Semantic SEO services: the topical map and naming work that gives every page one head phrase for its anchors to repeat.
Anchor text for image links, navigation, and over-optimization
How does alt text work as anchor text?
Google uses the alt attribute of a linked image as the anchor text of that link. A linked image with an empty or missing alt attribute is a link with no label, so the alt text of a linked image should name the target page in the same way a text anchor would, rather than describe the image’s appearance.
Do navigation anchors count as anchor text?
Navigation anchors count as anchor text, but they repeat on every page, so they label the site’s sections rather than describe a target in the context of one page. A navigation label such as “Technical SEO” sits among other section names, not inside a sentence about the target. Contextual anchors in body text carry the surrounding words that navigation lacks, which is why the two are written to the same names but do different work.
What is anchor text over-optimization?
Anchor text over-optimization is the repetition of exact-match commercial anchors in links placed to manipulate rankings, typically links built on other sites in volumes no independent linking would produce. Google’s Penguin update, launched in April 2012 and made part of the core algorithm in 2016, targeted this pattern, and Google’s spam policies treat link schemes as spam. Internal exact reuse is a different case: it gives each page one consistent name across the site’s own links, and it stays within that page’s name rather than inserting keyword strings into unrelated sentences.
Should anchor text ever be generic?
Anchor text should not be generic; Google’s link guidance names “click here” and “read more” as anchors to avoid. A generic anchor says nothing about its target to a search engine and nothing to a screen reader user who hears the link out of context, so the target’s name replaces it even in calls to action.
Anchors are assigned and capped when the content network is built. → Content network build