How to plan anchor text across a link profile
Plan anchor text for the whole profile, not link by link. Record the anchors each target page already has, compare them with the pages that rank, set a band for each anchor type per page, and give writers two or three variants. Google publishes no safe ratio, so every percentage is a rule of thumb.
In short
- Google publishes no safe anchor ratio, so a plan should be benchmarked against the pages that already rank for the same term and page type.
- Ahrefs' anchor text study of 384,614 pages, updated in April 2020, found only a weak positive correlation between exact-match anchor share and rankings, and cites an industry convention of roughly 1% to 5% exact match.
- One published guide for iGaming, from ICODA in September 2026, suggests 50% to 60% branded, 20% to 30% partial match, 5% to 10% exact match and 10% to 15% generic or URL anchors.
- Google's leaked documentation records the share of spam-phrase anchors and the days over which they arrived, according to 2024 analyses, so both the mix and the timing of commercial anchors are measured.
- The lowest-risk place for exact-match anchors is your own internal links, where you control every word.
Plan anchor text at the level of the whole profile, not one link at a time. Decide, for each page you build links to, roughly what share of its anchors should be branded, descriptive or keyword-led. Then choose each new anchor so the profile moves towards that picture. A single exact-match anchor is harmless. The same commercial phrase arriving from twenty weak sites in a month is the pattern that gets links discounted.
Google publishes no safe ratio. Every percentage in this lesson is an industry rule of thumb or one named guide’s suggestion, and is labelled that way. What anchor text is and how Google uses it is covered in its own lesson. This one is about the plan: how to find out where you stand, what to compare yourself with, and how to stop a run of reasonable individual decisions from adding up to a profile you would not have chosen.
Why plan at profile level
Two pieces of evidence explain the approach. Ahrefs’ anchor text study, updated in April 2020, looked at 384,614 pages across 19,840 keywords. It found only a weak positive correlation between the share of exact-match anchors and ranking, and most top pages had a low share. More keyword anchors did not reliably mean better rankings. That weakens the main argument for asking for the keyword every time, which is the belief that each exact-match anchor pushes the page up a little. On that data the pages at the top mostly got there with few of them.
Google’s Content Warehouse documentation, leaked in 2024, contains a group of fields that count spam-phrase anchors, record the fraction they make up and note the days over which most of them were discovered, according to iPullRank’s and Hobo’s 2024 analyses. The leak shows no weights. It does show that the share of commercial anchors and the speed at which they arrive are both recorded. So the upside of piling on keyword anchors is weak on the one study, and the downside is something Google keeps a count of. Neither of those can be seen by looking at one link. So the unit that matters is the distribution for a page, over time. That is what a plan controls.
Audit what you have, then benchmark
For each target page, export its anchors from a backlink tool. The backlink tools comparison lists products that provide this report. Then work through the export in order.
- Count by referring domain, not by raw link, so one sitewide link does not distort the picture.
- Put every anchor into one type: branded, naked URL, page title, generic, partial match, exact match, image or empty.
- Note the language of each anchor. English anchors pointing at a Greek or German page are a footprint in their own right.
- Record the totals for the page and for the whole domain.
Most sites find the homepage is heavily branded and a few commercial pages carry nearly all the keyword anchors. That is the starting position, and it explains why a domain-wide figure can mislead: a site can look comfortably branded overall while one category page has a profile made almost entirely of the phrase it wants to rank for. The per-page totals are the ones to act on.
Next, take the top five ranking pages for the term each target page competes on, and run the same count. Compare like with like: a homepage against homepages, a category page against category pages, a guide against guides. A link gap analysis uses the same set of competitors, so the two jobs can share one export. Expect the mix to vary by page type. Three patterns are what you will usually see. A homepage is dominated by the brand name and URL. A commercial or category page shows brand plus keyword, some partial match and few exact match. A guide, study or tool collects its page title and topical phrases, because writers cite it by name. Comparing a category page with a competitor’s homepage would therefore tell you nothing, since the two would differ even if both were entirely natural.
Natural profiles are dominated by brand, URL and title anchors. If the pages that rank show 3% exact match and yours shows 30%, the gap is your risk. If they show 3% and you show none, you have a little room. The benchmark works in both directions, which is its advantage over a fixed rule: it warns you when you are far beyond what the ranking pages carry, and it also tells you when a cautious profile could take a few chosen keyword anchors without standing out.
Set bands, not targets
Write down a range for each anchor type per page type. Use your benchmark first. Published heuristics are a second reference, and they disagree with each other, which is the point.
| Source | Branded and URL | Partial match or topical | Exact match | Generic |
|---|---|---|---|---|
| Industry convention cited in Ahrefs’ anchor study | Not stated | Not stated | Roughly 1% to 5% | Not stated |
| ICODA’s iGaming guide, September 2026 | 50% to 60% branded | 20% to 30% | 5% to 10%, as a hard limit per money page | 10% to 15%, including naked URLs |
| Rule of thumb reported among iGaming agencies | 60% to 70% | 15% to 20%, including generic | Under 10% | Counted with topical |
None of these comes from Google. Confidence in any numeric ratio is low. The exact-match column alone runs from roughly 1% to 5% in one row to a limit of 10% in the others, and the rows do not even divide the anchor types in the same way, so anyone presenting a single ratio as settled is choosing one of several opinions. Treat a band as a ceiling on keyword anchors, not as a quota to fill. If a page sits below the top of its band, that is a fine place to stay, and a plan that reads the band as a target will spend money pushing a healthy profile towards its limit. Competitive sectors such as those in our iGaming hub run closer to the upper end because every rival does.
Add two limits that are easier to enforce than percentages. The first is a cap on exact-match anchors per target page, counted in links, not in percent. The second is that no exact-match anchor is repeated from the same publisher. Percentages shift every time a new link arrives and need a fresh export to check, whereas a count and a no-repeat rule can be verified from the tracker in a moment by whoever approves the next order. They also hold on a page with few links, where a percentage swings wildly with each addition.
Build an anchor bank for each target page
For each page in the plan, write a short list of acceptable anchors by type: the brand, brand plus topic, two or three partial-match phrases, the page title and one exact-match phrase. Every option must describe the page honestly. An anchor that does not match its target is a waste, and the leaked documentation includes a demotion for anchors that do not match their target, though it shows nothing about how that is applied. The bank does its work at the moment an order is placed. Whoever writes the brief picks from a list that was agreed in advance with the whole profile in view, so the choice no longer depends on what seems appealing that day.
Deciding which pages go in the plan is a separate job, covered in which pages need links. A common working default from placement briefs sends 40% to 60% of new links to the homepage and brand pages, 20% to 30% to informational assets and 20% to 30% to commercial pages. That split is practitioner practice, not data, and it shapes anchors too: more homepage links means more brand anchors without any effort. The two plans are tied together in that way. If the page plan sends most links to commercial pages, the anchor plan is under strain from the start, because commercial pages are where the pull towards keyword anchors is strongest.
Decide where you control the anchor
You only choose the anchor for some links. Plan around that.
| Link source | Who chooses the anchor | Planning rule |
|---|---|---|
| Digital PR and editorial mentions | The journalist | Do not ask. Expect brand and URL anchors |
| Partner, supplier and directory links | Mostly the other site | Brand name |
| Genuine guest posts | You, within the host’s rules | Brand or a natural descriptive phrase |
| Paid placements | You | The only place a commercial anchor is a real choice, and where the cap applies |
| Internal links | You, fully | Use exact-match and descriptive anchors freely |
Earned links supply the branded base for free. That base is what makes a small number of chosen commercial anchors look ordinary. A site with only paid guest posts in its profile has no such cover, so its plan needs to lean branded even where it could choose otherwise. Put another way, the freedom to choose is greatest where the risk is greatest. Paid placements are the one external source where you set the words, and they can arrive in batches from sites that sell to others, so the restraint has to come from the plan.
The last row of the table is the outlet for everything the plan holds back elsewhere. On your own internal links you control every word, which makes them the lowest-risk place for exact-match anchors. When you do choose an anchor on someone else’s site, give the writer two or three variants and let the sentence decide. How to write a placement brief shows where this sits in the brief.
Log, pace and review
Add the anchor and its type to your link tracker for every new link. Without that record the plan cannot be checked, and the next audit starts again from a raw export. Then pace commercial anchors by spreading them across months. A burst of similar keyword anchors is what the leaked spike fields describe. Link velocity covers timing in general. A cap that is respected over a year can still be filled in a single fortnight, and on the evidence above it is the bunching, as much as the total, that is recorded.
Review quarterly. Compare the distribution per page with your bands and with the competitors again, since their profiles move too, and a benchmark taken a year ago describes a set of search results that may no longer exist. Require approval as well. A written rule that any commercial anchor needs sign-off stops a vendor from filling the cap in one order. That belongs in your link buying policy. Put the anchor bands into your link plan so they sit next to targets and budget, where the person deciding what to buy next will see them. Anchor types are defined in the glossary.
Common questions
What is a good anchor text ratio for backlinks?
There is no official one. Google has never published a ratio. The working method is to match the anchor mix of the pages that already rank for your term, and to keep exact-match anchors a small minority.
How do I analyse the anchor text of my backlinks?
Export the anchors report for each target page from a backlink tool, group every anchor into a type such as branded, URL, generic, partial match or exact match, and count by referring domain. Then do the same for the top ranking competitors.
Should every backlink use my target keyword?
No. A profile where most links use the same commercial phrase is the pattern Google's Penguin system was built to catch. Most anchors in a natural profile are the brand name, the URL or the page title.
How many anchor variants should I give a writer?
Two or three acceptable options are enough. Let the sentence decide which one fits, and never ask for an anchor that does not describe the page it points to.
How often should I review my anchor plan?
Check the distribution at least once a quarter and after any large campaign. Review it straight away if one commercial anchor starts to dominate the links to a single page.

