Anchor Text Ratios: A Practical Guide to Staying Safe

Anchor text is the visible, clickable words in a link. It’s one of the older ranking signals still doing real work, and it’s also the signal people most reliably ruin by trying too hard. Nearly every link-related penalty story from the last decade traces back to the same mistake: someone decided their target keyword was the anchor, and then used it forty times.

This guide covers the anchor types, gives example distributions for three different kinds of page, and explains why the ratio you obsess over is less important than the story your link profile tells about how it was built. It applies whether you’re earning links, running outreach, or buying a link insertion (also called a niche edit) inside an existing article. One thing to be clear about at the start — the numbers below are working targets drawn from how natural profiles tend to look. They aren’t the output of a study, and nobody outside Google knows the actual tolerances.

What is anchor text ratio, and why does it matter?

Anchor text ratio is the distribution of your backlink anchors across categories — how much of your profile is your brand name versus your target keyword versus a bare URL. It matters because Google has used anchor text as a relevance signal since the beginning, and because an unnatural distribution is one of the easiest manipulation footprints to detect at scale.

Think about how links happen when nobody is optimizing anything. A journalist cites you and writes your company name. A forum user pastes the URL. A blogger writes “this useful guide” and links it. Someone writes “according to a study by [Brand]”. Almost nobody, unprompted, writes “best commercial espresso machine for small cafes” as clickable text — that phrasing only appears when somebody was told to use it.

So the ratio isn’t a rule you’re obeying. It’s the residue of a natural process, and you’re trying to keep your deliberate link building from looking obviously unlike that process.

What are the main anchor text types?

There are six categories worth tracking. Most SEO tools will bucket them for you imperfectly, so it’s worth classifying a sample by hand at least once so you know what your tool is getting wrong.

Two categories get mishandled constantly. Branded-plus-keyword anchors (“SerpInsight link insertions”) are usually classified by tools as partial-match, and they’re genuinely useful — they carry keyword relevance while still reading like a normal reference to a company. And image links are anchors too. If every third-party site embedding your infographic uses the same keyword-stuffed alt attribute because you supplied the embed code, that’s an exact-match cluster you didn’t know you had.

What anchor mix should a commercial money page have?

Commercial pages carry the most risk because they’re the pages worth manipulating, so the target skews conservative. Branded and partial-match anchors should carry most of the load, with exact-match kept to a thin slice.

A worked example, hypothetically. Say you’re building 20 links to a service page over four months. That’s roughly 8-10 branded, 5-7 partial-match, 3-4 naked or generic, and 1-2 exact-match. Notice how small that last group is. Two exact-match anchors out of twenty feels almost pointless when you’re planning it, and it’s about right.

One more note on commercial pages: if the page already ranks on page two for its main term, the temptation to push with exact-match anchors is strongest exactly when the risk is highest. That’s the moment to add partial-match and internal links instead.

What about a homepage?

Homepages should be overwhelmingly branded, because that’s how homepages get linked in the wild. Anything above about 5% exact-match on a homepage looks strange, since almost nobody links to a company’s front door using a commercial keyword.

If your homepage profile doesn’t look roughly like this, the fix usually isn’t more links — it’s that your brand isn’t being mentioned enough anywhere. That’s a PR and content problem wearing an SEO costume.

What about an informational blog post?

Informational pages have the loosest constraints because they attract genuinely varied anchors. People link to a guide with descriptive phrases, titles, and generic references far more readily than they link to a pricing page.

Exact-match tolerance is higher here for an obvious reason: the exact-match anchor for an informational page is usually a natural sentence fragment. Someone writing “there’s a decent guide to anchor text ratios over here” has produced something that is technically an exact-match anchor and is also just normal English. Google can tell the difference between that and forty identical commercial phrases.

Treat all five of these tables as starting points you adjust against your own niche. In some verticals — local services especially — branded anchors dominate far more heavily than the tables suggest, because that’s how local businesses get mentioned.

Why do exact-match anchors cause trouble?

Because they were the defining footprint of the pre-2012 link-buying era, and Penguin was built to find them. Before that update, stuffing a page’s target keyword into hundreds of anchors was the fastest way to rank, and enough people did it that the pattern became trivially identifiable.

What made it detectable wasn’t just the percentage. It was the combination — the same phrase, repeated, on sites with no topical connection, appearing at a rate no organic process produces, often with the anchor sitting awkwardly in a sentence written around it. Any one of those on its own is unremarkable. Stacked, they describe a purchase order rather than a citation.

Penguin was folded into the core algorithm in 2016 and Google has said it now largely devalues manipulative links rather than penalising sites for them. In practice that changes the failure mode more than it changes the advice: the usual outcome of an over-optimized anchor profile today is that the links quietly stop counting and your money is gone, rather than a dramatic ranking collapse. That’s still a bad outcome. It’s just a quieter one.

Why does the acquisition pattern matter more than the ratio?

Because a profile is a time series, not a snapshot. Two sites can show identical anchor percentages today and look completely different in how they got there — and the “how” is what carries information about intent.

Consider two hypothetical sites, both landing at 8% exact-match. The first accumulated links steadily over three years across dozens of referring domains in varied niches, with exact-match anchors appearing occasionally and never twice in the same month. The second acquired nothing for two years, then took 45 links in six weeks, of which the exact-match ones all used the same four-word phrase and all appeared on sites that also sell links. Same percentage. Wildly different stories.

So when you plan anchors, plan across three dimensions rather than one:

• Variety within category. Ten partial-match anchors should be ten different phrasings, not one phrase repeated. “guide to anchor ratios”, “how anchor distribution works”, “this breakdown of anchor types” — all partial-match, all distinct.
• Pacing. Spread acquisition across months. A profile that gains links in a steady trickle reads differently from one that gains them in bursts followed by silence.
• Source diversity. Twenty links from twenty different kinds of site beats twenty links from twenty sites that all look like each other, run the same theme, and sell placements at the same price.
• Sentence fit. The anchor should read as if the writer chose it. If the paragraph only makes sense because the anchor was forced into it, a human reviewer will spot that instantly.
• Destination spread. If every link points at one money page, the anchor ratio on that page is the least of your problems. Real sites accumulate links across many URLs.

That last point is the one people underweight. A healthy profile has depth — links landing on blog posts, resources, tools, and the odd product page. Building depth is partly an internal problem too, and how you route authority internally between those pages often does more for a money page than another external link would.

What should you do if you’ve already over-optimized?

Dilute, don’t demolish. Add branded, generic, and naked-URL anchors over the following months until the aggressive ones shrink as a proportion, and resist the urge to email site owners asking them to strip links out.

Here’s the reasoning. Removing links removes whatever value they were still passing, and it does it immediately, whereas dilution reduces the proportion without giving anything up. Panic removal also tends to be indiscriminate — people pull good links along with bad ones because both used the same anchor. And if the links were already being algorithmically ignored, removing them accomplishes nothing except making your profile smaller.

A practical sequence:

1. Measure honestly

   Export your anchors per URL, not site-wide. Site-wide numbers hide the problem — a homepage full of branded anchors can mask a service page sitting at 40% exact-match.

2. Stop adding to the pile

   Freeze exact-match entirely on the affected URL. Not reduce. Freeze, until the proportion is back in a sane range.

3. Build branded and generic deliberately

   Digital PR, podcast appearances, supplier and partner pages, directory profiles, resource-page links. These are unglamorous and they’re what a real brand’s profile is mostly made of.

4. Add depth around the page

   Point new links at the blog posts and resources that link internally to the money page. The authority still flows, and the anchor pressure on the target URL doesn’t increase.

5. Only disavow with cause

   Disavow if you have a manual action or you inherited links from a genuinely toxic campaign. Otherwise leave it alone — the disavow file is a tool for damage, not a hygiene routine, and it’s easy to do harm with it.

Give it time. Anchor dilution is a months-long process, not a weekend fix, and the profile only moves as fast as you can add clean links to it.

How should you assign anchors across a batch of purchased links?

Decide the distribution before you buy anything, then assign each anchor to a specific placement based on how well it fits that page’s existing sentences. Anchor selection should be the last step of placement, not the first step of ordering.

That ordering matters more than it sounds. If you pick anchors first and then hunt for pages to put them on, you end up forcing phrases into paragraphs where they don’t belong. If you find the paragraph first and choose an anchor that suits the sentence already there, you get anchors that vary naturally because the sentences vary naturally. That’s the approach behind our own placements inside existing, already-ranking articles — the surrounding paragraph has to still read correctly with the link removed, which rules out most aggressive anchors before anyone has to make a judgement call about ratios.

Two supporting habits. Keep a simple spreadsheet of every anchor you’ve used per target URL, because you will otherwise repeat yourself. And vet the host site properly before you worry about the anchor at all — the twelve checks we run on a site before placing anything matter more than the wording of the link. An excellent anchor on a dead site is worth nothing, and it’s worth reading honestly about what Google actually penalises when it comes to bought links before you scale any of this.