Exact-match anchors are not poison, but the distribution around them matters more than the count.
SEO Mighty · 11 February 2026
Somewhere along the line the industry settled on a number. Keep exact match under a certain percentage and you are safe; go over it and you get filtered. The number changes depending on who is repeating it, which should be the first clue that nobody knows where it came from.
Ask for the source and you get a link to a blog post that links to another blog post. The chain never terminates in anything you can check. That does not make the underlying instinct wrong — over-optimised anchor profiles are a real thing, and there is a reason the advice persists — but a fixed threshold cannot be right for everyone, because the inputs differ enormously between sites.
Consider a brand whose name is also its main commercial term. A hosting company trading under the words "fast hosting" will accumulate that exact phrase from people who are simply naming the brand, and its profile will look aggressive by any percentage rule while being entirely natural. Now consider a brand with an invented name in the same niche. The identical anchor mix would look bought.
The other inputs that a fixed number ignores: the age and size of the profile, where the links sit, what the niche does by convention, and what the pages currently outranking you look like. A profile with a few dozen referring domains behaves nothing like one with thousands — a single anchor moves the distribution a long way in the first case and is invisible in the second.
The useful move is to stop thinking in percentages and start thinking in shapes.
A profile that accumulated on its own is dominated by people referring to the thing rather than describing it. That means brand names, bare URLs, the brand plus a word or two, and the page title as it was copied out of the source. It contains a long tail of one-off phrasings, because every person writing a link writes it slightly differently.
It also contains junk. "Click here", "this article", "read more", "here", the domain with the protocol still attached, a stray image link with no anchor at all. Real profiles are messy in ways that manufactured ones rarely are, and an over-managed profile is conspicuous for being tidier than anything that grew on its own.
Exact-match commercial anchors do appear in natural profiles. They are not poison. What is unusual is for them to appear in volume, all pointing at the same page, all phrased identically, all acquired in a narrow window. The count matters less than whether the surrounding distribution looks like something people wrote or something someone ordered.
The practical test: look at what already links to you, and ask whether the next anchor makes the shape more like a real profile or less like one.
Orders can be grouped into projects, and every project carries an anchor-text distribution chart built from the anchors in that project's briefs. It shows you what you are actually buying rather than what you meant to buy, which are often different once a campaign has been running for a few months.
The chart also raises an over-optimisation warning when the distribution concentrates too heavily on exact-match anchors. We are deliberate about what that warning is and is not. It is a prompt to look, not a verdict. It reads the anchors in the project, not your whole backlink profile — we do not have visibility into links you acquired elsewhere, and a project chart that looks aggressive on its own may be a small part of something balanced.
Treat it as the thing that catches the case nobody meant to create: three placements ordered weeks apart, each one reasonable in isolation, that turn out to use the same commercial phrase pointing at the same page.
Decide the anchor for each placement against the profile the target page already has, not against a table. That is more work than applying a rule, and it is the only version of this that survives contact with a real site.
If a page has one inbound link, an exact-match commercial anchor should not be its only signal. Give it the brand, the URL, or the title first, and earn the right to be specific later.
Where you do want commercial relevance, vary the partial rather than repeating one string. A phrase with the brand attached, a longer question-shaped version, the term with a qualifier — these carry the same topical meaning without stacking identical anchors. Repeating one exact string across placements is the pattern worth avoiding, more than the raw count of exact matches.
Anchor text goes in the brief along with the target URL, and we validate it for length, embedded URLs and spam patterns. That validation is a floor, not editorial judgement — it will stop something malformed, it will not tell you your anchor strategy is wrong. Where a placement is paid we support rel="sponsored", and the link type is stated on every listing, so the anchor is not carrying a claim about the link that the markup contradicts.
If you want a second opinion before the brief is finalised, ask us. We will say when an anchor looks heavier than the page can carry, and we would rather have that conversation before the article is written than after it is live.