Anchor text is the part of link building where enthusiasm does the most damage. It is the one variable a buyer directly controls, so it is the one they over-control — and a profile where forty per cent of anchors are the exact phrase someone wants to rank for is not a profile, it is a confession.
Why it goes wrong
The logic is seductive. If you want to rank for "commercial roofing Nottingham", surely you want links saying "commercial roofing Nottingham"?
Directionally, yes. The problem is that naturally acquired links almost never look like that. Real people linking to a business use its name, or the URL, or a phrase like "this roofing company" or "as covered here". Keyword-rich anchors happen, but as a minority.
So a profile dominated by exact-match anchors is not just aggressive, it is statistically anomalous — and anomaly detection is a much easier engineering problem than judging content quality. It is one of the cheapest signals available.
The distribution we build to
A workable starting point for most UK businesses:
| Anchor type | Target share | Example |
|---|---|---|
| Branded | 40–50% | "Vesta Blinds", "Vesta Blinds Ltd" |
| Partial match | 20–25% | "blinds fitted across Nottingham" |
| Generic | 10–15% | "this guide", "read more", "here" |
| Naked URL | 10–15% | "vestablinds.co.uk" |
| Exact match | 5–10% | "blinds Nottingham" |
Two caveats. First, these are targets for the profile as a whole, not a rule for each individual link. Second, they are a starting point — the right mix depends on what you already have, which is why an audit comes before a plan.
Recency matters more than the lifetime average
This is the part most guides miss. A site with 300 links built over eight years, of which 8% are exact match, looks fine. The same site adding twelve links this quarter, ten of them exact match, does not — regardless of what the lifetime percentage says.
Search engines can look at your profile over time, and a sudden shift in anchor composition is a stronger signal than a static ratio. In practice this means the first placements in any campaign should be branded, establishing the baseline, before anything keyword-rich enters the mix.
Nobody naturally links to a business with the exact phrase it wants to rank for, ten times, in a fortnight. That is the entire detection problem, and it is not a hard one.
How to audit what you have
Twenty minutes with any backlink tool:
- Export every anchor pointing at your domain.
- Bucket them into the five categories above. Do it manually if the list is under a few hundred — the edge cases are the interesting ones.
- Calculate the percentages and compare to the table.
- Now do it again for the last twelve months only. This is the number that matters.
- Repeat per page. A page with six links, five of them exact match, is a problem even if the domain-level profile looks healthy.
That last step catches more issues than the domain-level view. Over-optimisation happens at page level, because that is where campaigns are aimed.
If you are already over-optimised
Common, and usually fixable.
Stop immediately. No more keyword anchors until the ratio recovers. This is the whole first phase.
Dilute rather than remove. Build branded, naked-URL and generic anchors over the following months. Dilution is faster, cheaper and lower-risk than trying to get links removed — and removal requests to sites you paid rarely go well.
Only disavow genuine junk. If the links themselves come from deindexed or obviously spammy sites, that is a different problem and disavow is appropriate. Over-optimised anchors on decent sites are a dilution problem, not a removal one.
Give it a quarter. Recovery is gradual. Chasing it with more link buying is how people turn a minor issue into a real one.
Practical rules we apply
- First two or three placements in any campaign: branded anchors only
- Never the same exact-match anchor twice in a row
- Anchors should read like part of a sentence, because they are
- Vary the target page — sending everything to one URL is its own footprint
- If a client insists on ten exact-match anchors, we say no and explain why
That last one costs us the occasional sale. It costs a lot less than the alternative, which is a client who spent six months getting worse and correctly blames us for it.
We publish the same table on the backlinks page rather than treating it as proprietary, because it is not — the value is in applying it consistently, not in knowing it exists.
Frequently asked
What percentage of anchors should be exact match?
For most sites, somewhere between five and ten per cent of the total profile. The figure matters less than the direction of travel — a site whose recent links are almost all exact match looks manipulated regardless of what the lifetime average says.
Do branded anchors help rankings?
Yes, both directly and indirectly. They pass equity like any other link, and they establish the natural-looking baseline that lets a smaller number of keyword-rich anchors do their job without the profile looking engineered.
How do I fix an over-optimised anchor profile?
Stop buying keyword anchors immediately, then dilute with branded, naked-URL and generic anchors over several months. Removing existing links is usually harder and less effective than burying them, unless the host sites are themselves problematic.
Want this done rather than explained? We own 38 UK publications and place editorial articles and links on them from £65. Send the target page and we'll come back within one working day with a plan.