Anchor text

Anchor text is the clickable wording of a link. Google uses it as a core relevance signal, and heavy keyword anchors are also the classic sign of link spam. No safe ratio of anchor types has a primary source.

In short

  • Anchors are one of three core relevance signals in Google, alongside page body text and user clicks, according to an exhibit from the US antitrust case reported in May 2025.
  • Google publishes no safe ratio of anchor types, so every percentage you see quoted is an industry rule of thumb.
  • Ahrefs' anchor text study of 384,614 pages found only a weak positive correlation between exact-match anchor share and rankings, and most top pages had a low share.
  • Google's spam policies name articles, guest posts and press releases with optimised anchor text as link spam.
  • The working method is to compare your anchor mix with the pages that already rank for the same term, and to keep keyword anchors mainly for your own internal links.

Anchor text is the clickable wording of a link. In the sentence “read our guide to tyre pressure”, the words you can click are the anchor. Search engines read those words as a description of the page the link points to, which is a sensible thing for them to do: the person who wrote the link was summarising the target in a few words for their own readers, and a summary written by an outsider is harder to fake than anything the page says about itself.

That is what makes anchors one of the most useful parts of a link and also the easiest part to overdo. If the wording of a link tells Google what a page is about, the temptation is to make every link say the keyword you want to rank for, and that temptation is exactly what Google’s link spam systems were built around. The rest of this lesson sets out the types of anchor, what is known about how Google treats them, and where that knowledge runs out, because a good deal of the confident advice on this subject rests on nothing firmer than habit.

Types of anchor text

The names for anchor types are easier to follow with an example, so take a fictional shop called Northwind Cycles with a page that targets “road bike tyres”. Every link to that page falls into one of the groups below, and the same link can sit on the border between two of them.

Type Example anchor Notes
Branded Northwind Cycles The most common type in natural profiles
Naked URL northwindcycles.example The address itself as the link
Generic click here, this site Says nothing about the target
Exact match road bike tyres Identical to the target keyword
Partial match guide to choosing tyres for a road bike Contains the keyword inside a longer phrase
Brand plus keyword Northwind Cycles road tyres A mix of the two
Page title Road bike tyres: sizes and pressures explained Common when writers cite an article
Image (the image’s alt text) Alt text stands in for the anchor

Natural profiles are dominated by brand, URL and title anchors, and the reason is plain once you think about how people write. Someone recommending a shop names the shop. Someone citing an article pastes its title or its address. Very few people, left to themselves, link to a business with the bare commercial phrase that business hopes to rank for. The mix also shifts with the kind of page. Homepages attract mostly branded anchors, because a link to a homepage is a reference to the company as a whole. Deep informational pages attract more topical and title anchors, because writers tend to cite an article by what it says. That difference matters later, when you come to compare your own anchors with anyone else’s.

What is known about how Google uses anchors

The firmest single piece of evidence is that anchors are a core relevance signal. An exhibit from the US antitrust case against Google, reported in May 2025, names three signals that make up topical relevance: anchors, body and clicks. This carries more weight than the usual public statements because it comes from a Google engineer under legal process, where vague reassurance is not an option. It places the words other sites use to describe your page alongside the words on the page itself and the behaviour of searchers, so anchors are a central input to how Google decides what a page is about.

The words around the link count too. The Google documentation leaked in 2024 includes fields that store the text to the left and right of a link, and the font size of the anchor. The leak lists data fields without weights, so it shows what Google stores and gives no indication of how much any of it matters. The same leak lists a demotion for anchors that do not match their target. Its description is minimal, and analysts read it as a check on whether the anchor fits the page it points to. Taken together, these suggest that a keyword anchor dropped into an unrelated sentence, pointing at a page about something else, is the kind of link the system is able to recognise as out of place.

Google also appears to measure spikes of spammy anchors. The leaked anchor fields include counts of spam-phrase anchors, the period over which they appeared and how many were demoted. Analysts read this as evidence that Google looks at the share and the rate of commercial anchors, and that anchors inside a sudden spike can be discounted. Some caution is needed here, because the exact field names come from analyses of the leak, and whether they are in live use is not confirmed. What it does tell a practitioner is that timing may matter as well as proportion: a batch of identical commercial anchors arriving in a short period could be treated differently from the same anchors gathered slowly.

On the policy side there is no ambiguity. Google’s spam policies, last updated in August 2026, list “articles, guest posts or press releases with optimised anchor text distributed on other sites” among their examples of link spam. Forum comments with optimised links are on the same list. So the one place where Google names anchor text in its written rules is the case where the site owner chose the wording and placed it on other people’s sites in bulk.

Against all that sits the question of how much keyword anchors help in the first place. Ahrefs studied the anchors of 384,614 pages across 19,840 keywords. The share of exact-match anchors showed only a weak positive correlation with ranking position, and most top-ranking pages had a low share of exact-match anchors. Partial-match and phrase-match anchors behaved in a similar way. The reading that follows is that keyword anchors are not a strong lever on their own, and that the pages which win are not winning through anchor stuffing. Put next to the policy wording, that makes the trade a poor one: the upside of pushing exact-match anchors looks small in the data, while the downside is written into the spam rules.

Why no safe ratio exists

You will see anchor text ratios quoted with confidence: so much branded, so much exact match, sometimes laid out to the nearest percentage point. None of them has a primary source. Google publishes no safe ratio. The Ahrefs study cites an industry convention of keeping exact match to roughly 1% to 5%, and describes it as a convention, which is the honest label for it. Quoted without that label the figure sounds like a rule, but a convention is all it is.

There are three reasons to distrust any universal figure. The first is that what looks natural differs by page type: a homepage and a guide attract different anchors, so a single ratio applied to both would be wrong for at least one of them. The second is that it differs by market. In some markets every ranking site has aggressive anchors, and in others almost none do, which means a profile that blends in among one set of competitors would stand out sharply among another. The third is that Google appears to look at patterns over time, such as spikes, and not only at a static percentage, so a profile could sit inside a quoted ratio and still show the pattern the leaked fields describe.

For those reasons our confidence in any numeric ratio is low, and we do not publish one. See the methodology for how we treat claims like this. The practical consequence is that the question “what percentage is safe” has to be replaced by a comparison with real pages in your own market, which is slower than reading a number off a chart but is the only version of the exercise with evidence behind it.

A working method

Since no universal figure can be trusted, the method is to measure what is normal among the pages you are competing with and to stay within it. It can be done with any backlink tool, and it is worth repeating for each important page, because the answer for a homepage will not hold for a guide.

  1. Benchmark the pages that rank. Take the top five results for your target term and look at the anchors pointing to the equivalent page on each site. Use the same page type: homepage against homepage, guide against guide.
  2. Note the range, not an average. You are looking for what is normal in that market, and for how far the most aggressive ranking page goes.
  3. Stay inside the range. If no ranking page has more than a handful of exact-match anchors, a profile full of them will stand out.
  4. Let most anchors happen. When a journalist or blogger links to you unprompted, they choose the words. That variety is what a natural profile looks like.
  5. Use your own site for keyword anchors. Internal links are where exact-match anchors carry far less risk, and you control every one.

The range matters more than the average because an average hides the outliers, and the outliers are what you need to see. If one ranking page carries far more keyword anchors than the rest, that tells you how much the market tolerates. It does not tell you that copying it is wise, since that page may be ranking in spite of its anchors. The last step deserves more attention than it gets. Internal links let you describe your own pages in the exact words you want, and because nobody is surprised when a site labels its own pages clearly, that is where the keyword work belongs.

Where you control the anchor, and what happens when it goes wrong

You control anchors in internal links, in your own author bio on a guest article and in listings you fill in yourself. You can also suggest one when you buy a placement, and that last case is where the risk concentrates. A paid article on a weak site with one exact-match commercial anchor is the pattern named in Google’s spam policy. If you buy placements, vary the wording, favour brand and descriptive anchors, and make the sentence read as though an editor wrote it. Anchors in another language need the same care. A keyword anchor in one language on a site written in another stands out to a reader, and it is reasonable to assume it stands out to a search engine.

Google’s Penguin system, launched in 2012 and part of the core algorithm since 2016, is the one built around manipulative links and anchors. The leak suggests its logic drops anchors instead of punishing the whole page. In practice that means over-optimised anchors are more often ignored than penalised, so the likelier cost of overdoing it is money spent on links that count for nothing, with no warning that this has happened. That is not the whole of the risk, though. A profile dominated by such anchors can still draw a manual action. Google link penalties and recovery covers what to do then.

Where to go next

Anchor text is only one of the features Google reads from a link, and it makes most sense alongside the others, so read how search engines evaluate links next. After that, guest posting and niche edits are the tactics where you are most likely to be asked which anchor you want, so the reasoning in this lesson applies most directly there.

Common questions

What is anchor text?

It is the visible, clickable wording of a link. For an image link, the image's alt text does the same job.

What is a safe anchor text ratio?

Nobody outside Google knows, and Google has never published one. The common advice to keep exact-match anchors to roughly 1% to 5% is an industry convention, not a rule from a primary source.

Is exact-match anchor text bad?

Not in itself. A few exact-match anchors are normal. A profile where many links from weak sites use the same commercial phrase is the pattern Google's Penguin system was built to catch.

Which anchor should I ask for when I earn a link?

In most cases, let the writer choose. When you do have a say, your brand name or a plain description of the page reads naturally and carries the least risk.

Vendors to look at

  • BazoomEditor's pick

    Sponsored content and link marketplace, managed service

  • MotherlinkEditor's pick

    Backlink services, guest posts, niche edits, full SEO