Discussion
  • Entity SEO
  • Internal Linking

Linking by entity rather than exact-match keyword: has anyone tested entity-based internal linking against keyword anchors?

Published 1 reply

Discussion starter: an entity SEO question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.

Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.

1 reply

  • admin
    Admin
    Original poster

    I am not aware of a published, controlled test that pits entity-based internal linking against exact-match keyword anchors, and the comparison is harder to run than it sounds, because the two approaches usually differ in how many links get added and where, not just in the anchor wording. What the available evidence and Google’s own guidance support is this: descriptive, natural anchor text is enough, and link placement and consistency matter more than whether the anchor repeats the exact target phrase. “Entity-based linking”, done well, is really a discipline about link targets: every mention of a concept links to the one page that owns it, using whatever natural phrase the sentence contains.

    Separate the two ideas

    The question bundles two different choices.

    • Which page gets the link. Entity-based linking maps each entity (a product, a condition, a location, a concept) to one canonical page, and links mentions to it consistently. Keyword-based linking often maps keywords to pages, which drifts when several pages target overlapping terms.
    • What the anchor says. Entity-based anchors use the entity name as it appears in the sentence, including variants and synonyms. Exact-match anchors repeat the target keyword, sometimes by rewriting sentences to fit it.

    The first choice has a clear mechanism: consistent targets reduce ambiguity about which page is about what, and help avoid internal cannibalisation. The second has much weaker evidence either way.

    What Google says about anchors

    Google’s link best practices ask for anchor text that is descriptive, reasonably concise and relevant to the page it links to, and warn against stuffing keywords into anchors. They do not say exact-match internal anchors are harmful, and they do not say varied anchors are better. The reading I would take is that Google wants the anchor to tell users and crawlers what the linked page is about, which both approaches can do.

    Why a clean test is difficult

    • Adding entity links usually increases the number of internal links to target pages, which is a bigger lever than anchor wording.
    • Entity linking changes which pages get links, sometimes resolving cannibalisation, again a bigger effect than wording.
    • Anchor text effects, if any, are small and slow, while rankings are noisy.

    So a test that compares “entity linking” with “keyword linking” usually measures link volume and targeting, not anchor style. If you want to isolate anchor wording, you need to hold the number of links, their placement and their targets constant, and change only the words.

    How I would design the test

    1. Pick a set of target pages of the same type, for example 200 product guides, each with a clear primary query.
    2. Add the same number of contextual links to each page, from comparable source pages, in the same positions.
    3. Randomly assign anchor style: half get exact-match anchors, half get natural entity anchors (the entity name as it appears, with variants). Keep a third group with no new links as a baseline if you can.
    4. Measure clicks and impressions for the primary query and for the wider query set for each target page, after the source pages have been recrawled.
    5. Read the result against the baseline. If both linked groups improve similarly, anchor style did not matter; the links did.

    What I would expect:

    Both linked groups to outperform the baseline, with little difference between them. If exact-match anchors show a small edge on the exact query, natural anchors may show a small edge on the wider query set, because variants cover more of how people search. That is a hypothesis, not a result, but it matches how anchors are described in Google’s documentation.

    Where exact-match anchors make sense anyway

    There are places where the exact target phrase is simply the right label, and forcing variety there would make the site worse. Navigation menus, breadcrumbs, category links in product listings and “related” modules usually use the page’s name, which is often its main keyword. That is normal and expected. The concern about over-optimised anchors comes mainly from external link schemes, where many sites linking with the same commercial phrase is a manipulation signal. Internally, the risk is more about readability: sentences rewritten to fit an awkward keyword, or every mention of a term turned into a link. If an anchor reads naturally to a person, it is fine for search.

    How to implement entity-based linking well

    • Build an entity map: each entity, its name variants and the one URL that owns it. This is the real work and it pays off beyond linking.
    • Link the first meaningful mention in a piece of content, not every mention.
    • Use the phrase that fits the sentence. Do not rewrite sentences to insert the exact keyword.
    • Handle ambiguous terms (homonyms, product names that are common words) with manual review or context rules, or auto-linking will create wrong links.
    • Exclude links from boilerplate and from pages whose topic is unrelated.
    • Audit the result: count inbound internal links per entity page and look for pages that now receive links for concepts they do not cover.

    Watch for this:

    Automated entity linkers can create hundreds of links per page on content-heavy sites. Beyond a point, links stop helping users and dilute each other. Cap links per page, and check a sample of pages as a reader would see them.

    Hypothetical example:

    A health publisher has three articles targeting variations of “iron deficiency anaemia symptoms” because different writers built links to whichever page they knew. The team creates an entity map, names one page as the owner of the condition, and relinks all mentions across 1,500 articles to that page using the natural phrases in each sentence. The owner page gains visibility and the two competing pages are merged into it. The improvement comes from consolidation and consistent targeting; the anchor wording was never the variable that mattered.

    Adopt entity-based linking for its targeting discipline, write anchors that describe the linked page naturally, and do not spend testing budget on anchor wording until you have fixed which pages receive links. If you do test wording, hold everything else constant, or you will measure something else.

    Need help with this on your own site?

    Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.

    See SEO services

Add a reply

Replies are open to approved members. Log in or apply to join.