The short answer: there is no magic anchor text ratio
For most external link profiles, exact-match anchor text should sit in the low single digits as a share of the total, with branded and naked-URL anchors doing most of the work. But the number is not the point. Google does not check your exact-match percentage against a fixed limit and penalize you the moment you cross it. It flags anchor text patterns that could not have happened naturally for a page of your size and link count.
That distinction changes everything. A 15% exact-match ratio is genuinely risky for a page with 40 referring links and effectively invisible for a page with 4,000. The same percentage means "someone clearly built these on purpose" in the first case and "statistical noise" in the second. So the useful question is not "what ratio is safe?" It is "does this profile look like something nobody engineered?"
This piece is a spoke of our complete guide to link building, and it deals with the external anchors you can influence. Internal anchors follow different rules, and we separate them clearly below.
What anchor text is, and why Google cares
Anchor text is the clickable words that carry a link. Google uses it as a signal about what the destination page is about, which is why Google's own guidance on descriptive link text tells you to write anchors that are specific enough to make sense out of context. Its self-test is blunt: read the anchor text alone and check whether it still tells you what the page is about.
The problem is that this signal is easy to abuse. If forty different sites all link to your page with the exact phrase "cheap car insurance quotes," that isn't a coincidence. Real editorial links produce messy, varied anchors: your brand name, the bare URL, "this study," "read more here," a half-sentence of context. When the anchors are too clean, too keyword-rich, and too consistent, they read as manufactured. Google's link spam policies name "links with optimized anchor text in articles, guest posts, or press releases" as a link-scheme signal, and illustrate it with the classic forum-signature example stuffed with anchors like "best pizza san diego."
The types of anchor text
You need the categories before you can think about distribution. The common types:
- Branded — your company or site name (for example, "DeployFlare"). The safest and usually the most common in a natural profile.
- Naked URL — the raw link, like deployflare.com. Extremely common in real citations.
- Exact match — the target keyword verbatim ("anchor text optimization"). The highest-risk type at volume.
- Partial match — the keyword inside a longer phrase ("a guide to anchor text ratios"). Lower risk than exact match, still keyword-bearing.
- Generic — "click here," "this page," "read more." Natural filler that shows up everywhere.
- Image links — where the alt text becomes the anchor.
- Related / topical — synonyms and adjacent terms that aren't your money keyword.
A profile built mostly of branded, naked-URL, and generic anchors is what unmanaged, organic linking looks like. Ahrefs studied 384,614 pages and found that the median share of exact-match anchors was zero across every ranking position, and that the correlation between exact-match anchors and rankings was weak (a Spearman coefficient around 0.14). In other words, top-ranking pages are not the ones stuffed with keyword anchors, and chasing a target ratio is optimizing the wrong variable.
A distribution that looks like nobody engineered it
Here is a rough shape for a typical external profile. Treat it as a sanity check, not a target you optimize toward.
| Anchor type | Rough share | Why |
|---|---|---|
| Branded | 35–50% | People cite you by name; this is the natural default |
| Naked URL | 15–25% | Bare links are everywhere in real citations |
| Generic | 10–20% | "here," "this article," "source" |
| Partial / topical | 10–20% | Contextual mentions that touch your topic |
| Exact match | 1–5% | The riskiest slice; keep it small |
If your branded and URL anchors together clear roughly half your profile, you are usually fine. The moment exact-match creeps into double digits on a page with only a few dozen referring domains, you've built a footprint. On a page with thousands of organic links, the same absolute count of exact-match anchors vanishes into the noise, which is exactly why large, established sites can carry distributions that would sink a small one.
That is the statistical-implausibility idea in practice. There is no universal threshold because the threshold scales with your size. A concentration of exact-match anchors becomes a problem when it is higher than ordinary editorial linking could plausibly produce for a page your size.
Can anchor text cause a penalty?
Yes, in two ways, and it helps to separate them.
The first is an algorithmic discount. Google's link-spam systems, powered by SpamBrain, can simply stop counting links whose anchors look manipulated. Google's March 2024 spam updates documentation describes this AI-based detection of link spam directly. You don't get a message; the links just stop helping, and the effort was wasted.
The second is a manual action for unnatural links, where a human reviewer flags the profile and rankings drop until you clean it up. Google's manual actions report documentation lists "unnatural links to your site" as a specific action type and points you to a reconsideration request once the profile is cleaned. Both routes trace back to the same root cause: a pattern that doesn't occur naturally. If you suspect anchor-related ranking damage, work through a structured backlink audit process rather than guessing, and read our guide to identifying and removing toxic backlinks for the cleanup mechanics.
One practitioner note worth stating plainly: over-optimized anchors are usually a symptom, not the disease. If you are anxious about exact-match ratios, it often means the underlying links were built at scale from thin sites. Fix the acquisition strategy and the ratio problem tends to solve itself.
Internal anchor text: different rules entirely
Everything above concerns links you don't fully control. Internal links are yours. You choose every anchor on your own site, which means the over-optimization worry mostly evaporates and the advice flips.
For internal links, be descriptive and specific. If you are linking to a page about keyword research, "keyword research" or "how to find low-competition keywords" is a better anchor than "click here." You are helping Google understand your site structure and passing relevance to the right page. You cannot easily create a suspicious pattern here, because a site linking to its own pages with clear anchors is exactly what Google wants to see. The mechanics of doing this well live in our guide to internal linking for SEO.
The one caveat: don't point every internal link at one page with the identical exact-match anchor and nothing else. Even internally, robotic uniformity looks off. Vary it the way you would write for a reader.
How to make external anchors look natural
You mostly can't dictate the anchors other sites use, which is the point. But you influence them through how you earn links.
- Earn links across different methods. A profile fed by guest posting, digital PR coverage, and broken link building naturally produces varied anchors, because each channel links differently. Journalists cite your brand; resource pages use your title; bloggers drop naked URLs.
- Let editorial links stay editorial. When someone links because your content is genuinely useful, they choose their own words. That messiness is the safety.
- Suggest brand or URL anchors when you do have influence. In guest posts or partnerships where you can request an anchor, default to branded or partial-match, not exact-match.
- Watch the whole profile, not one link. Anchor health is a distribution property. Check it in aggregate. The broader tactics sit in our guide to how to get backlinks, and teams working with thin profiles should read link building for startups, where every new anchor moves the ratio more.
Auditing your anchor distribution
Pull your referring domains and group anchors by the categories above. Two things matter together: the shape of the distribution and the quality of the domains behind it. A clean-looking ratio built on spammy sites is still a liability, and a slightly keyword-heavy ratio on strong editorial domains is often fine.
You can run this inside DeployFlare's backlink analysis, which groups anchors by type and flags concentration, so you see whether your exact-match slice is a plausible few percent or an engineered footprint. Because DeployFlare is priced from ₹499/month with UPI and GST billing, that check costs a fraction of what the enterprise suites charge for the same data. Comparing against rivals through competitor backlink analysis is also grounding: it shows you what an organic profile in your niche actually looks like, so you calibrate to reality instead of a made-up rule. This matters across markets, whether you rank in English, Hindi, or Tamil SERPs, since a natural anchor profile in one language rarely mirrors another.
Two habits keep you safe long-term. First, review anchor distribution every quarter, not just when rankings drop, because footprints build gradually. Second, judge new links by whether they would exist if Google didn't. If a link only makes sense as an SEO play, its anchor is probably part of a pattern you'll regret. Run the numbers in a full site and backlink audit and treat any anchor type climbing toward implausibility as the early warning it is.