Anchor Text Audit — Diversity Score + Penguin Risk

A natural backlink profile is mostly messy

When people link to you without being asked, they use your brand name, your bare URL, the page title, or something vague — "click here", "this article", "here". Exact-match commercial phrases turn up occasionally and largely by accident.

Which inverts the intuition: a profile dominated by exact-match keyword anchors is not a well-optimised profile. It is a profile that looks built, because that distribution does not arise on its own. The tidiness is the signal, not the optimisation.

What this tool does

Paste a list of anchors, one per line, and it sorts them into six categories — exact-match, partial-match, branded, generic, full URL and naked domain — and returns a diversity score.

Be clear about what that score is: it is 100 − (exact-match% × 2). A single linear transform of one number. It is a compact way to read how exact-match-heavy your list is, and nothing more — it is not a risk assessment, and no threshold in it corresponds to any published rule.

The six categories

• Exact-match — your target keyword precisely. The type that accumulates mainly when links are being built deliberately.
• Partial-match — the keyword with other words around it. Arises naturally when someone describes your page in a sentence.
• Branded — your site or company name. The natural default for most sites.
• Generic — "click here", "read more", "this page" and similar. Extremely common in real linking and often treated as a flaw when it is simply how people write.
• Full URL and naked domain — what people paste when they cannot be bothered to write anything.

What the tool cannot tell you

Three things, and they matter more than the distribution:

It has no backlink data. It analyses the list you paste and cannot discover your links. Export them from whichever backlink tool you use — Ahrefs, Semrush, Moz, Search Console — strip the header row, and paste one anchor per line.

It knows nothing about the linking sites, and their quality and relevance matter far more than the words in the anchor. A varied anchor profile built from worthless sites is not a good profile.

There is no published threshold. Ratios quoted in SEO writing are inferences, not documented rules, and Google folded its dedicated link-spam filtering into core ranking years ago rather than running it as a separate periodic check. Treat any specific percentage presented as a safe limit with suspicion, including ones you may read elsewhere.

What to do about a skewed distribution

Almost never remove links. If exact-match anchors dominate, the fix is to earn more natural links so the proportion falls — the numerator stays put and the denominator grows.

Disavowing links you did not place is rarely the right first move, and doing it in haste removes links that were helping. If you did build the links deliberately, the useful change is to how you approach future outreach rather than an attempt to unwind the past. And if the pattern arose without your involvement, it is usually not something to act on at all.

Analysing an existing profile tells you what to aim for next. To generate anchor variations that move the distribution rather than reinforce it, use the anchor text builder.