Anchormap – AI Internal Linking & Link Map

Yes. Internal links help search engines discover and crawl your pages, understand which pages are related, and decide which pages are most important. They also spread authority from strong pages to newer ones and keep visitors on your site longer. Anchormap makes finding and adding them fast.

A page with no internal links pointing to it from the rest of your site. Search engines may crawl it rarely or not at all, and it receives no internal authority. Anchormap marks orphan pages with a red ring on the map and lets you filter the table to show only orphans.

There is no fixed number. Add a link wherever it genuinely helps the reader: typically a handful in a normal-length article, more in long guides. The Max suggestions per page setting keeps Anchormap's suggestions focused.

No. Without one, matching uses TF-IDF keyword similarity over your own content and everything still works. An embedding model makes the matching semantic rather than keyword-based.

No. Links are only written when you press "Insert link" on a specific row, one row at a time. Each insertion creates a post revision, so it can be undone from the editor.

No. When the model suggests different anchor text, it is used only if that exact string already exists in the source post. The plugin wraps existing text in a link; it never writes new text.

Inside an existing link, a heading, a code or preformatted block, a blockquote, a shortcode, or a block comment. Only body prose is considered.

Anchormap works alongside any SEO plugin and any theme; it only reads and edits post content. Content that lives in the normal post content (the block editor and the classic editor) is fully supported. Page builders that store text elsewhere, or blocks that keep their text in block attributes, leave little body text to link from, so those pages still appear on the map but may get fewer suggestions.

Yes. Any public post type can be enabled in the settings.

Both, and it matters which half does what. Which pages are related is decided by AI (embeddings), or by keyword similarity when AI is off. Which words become the anchor is decided by a scoring heuristic, not AI. Whether a specific link is worth making is decided by AI only when you press "Ask AI" on that row.

0.55 × similarity + 0.45 × anchor quality, plus a bonus when the destination has no inbound links. Similarity is the cosine similarity between the two pages, rescaled against your own site's distribution on every scan. Anchor quality combines phrase length, word rarity, overlap with the destination's title weighted by how distinctive the shared words are, and a check that the destination really is the page most about that phrase.

It sends one row to your chat model with the anchor, the sentence around it and a summary of the destination, and asks whether the link helps the reader. It adopts a different anchor only if that exact text already exists in your post, blends its confidence into the score rather than replacing it, and explains its reasoning under the anchor. It costs a fraction of a cent per press, and nothing until you press it.

Because one-word anchors like "services", "analysis" or "reports" match a related page but tell the reader nothing about where they are going. Single words are rejected unless the destination's whole title is one word. The anchormap_min_anchor_words filter overrides this.

Yes, including when formatting splits the phrase: AI visibility tracking is matched and linked as a whole. A phrase is skipped only when wrapping it would split a tag pair and produce invalid HTML, or when it sits in a heading, an existing link or a code block.

No. Anchormap only runs in the admin when you start a scan, in small batches. It adds nothing to your front end: no scripts, no styles and no markup for visitors.