Internal Linking

Internal Linking analyzes your article inventory and suggests sensible cross-links — anchor text, target article, position, and rationale included. Instead of hunting through the editor for which old article fits the current topic, you get a sorted list with a confirm button. Every accept or reject trains the ranking; the more you use it, the sharper the suggestions. Mandatory tool for anyone wanting to weave a topic cluster cleanly together without tracking link-building manually.

What it can do

• Per-article suggestions — for every article, a list of matching link targets from your own inventory, with anchor, source sentence, and confidence score.
• Project-wide analysis — a single run scans all articles in a project, builds the semantic graph, and fills the suggestion queue.
• Approve / Reject / Edit — per suggestion, you decide whether the link is set, rejected, or applied with a different anchor.
• Editor integration — confirmed links are patched directly into the article HTML; no copy-paste.
• Status tracking — pending / approved / rejected / applied with filters so you don't process anything twice.
• Anchor diversity — the suggestion algorithm makes sure the same target article isn't linked 20 times with the same anchor.

When to use

• You have 20+ articles in a project and lose track of what links where.
• You're building a topic cluster and need pillar → cluster → pillar interlinking.
• You want to give old articles fresh traffic by having new ones link to them.
• You want to keep anchor-text profiles clean instead of writing "click here" 50 times.

Workflow

1. Start project analysis — POST /projects/{project}/internal-links/analyze or the UI button. Job runs in the background (see Automation).
2. Review suggestions — GET /articles/{id}/link-suggestions lists the top suggestions per article.
3. Decide — PUT /link-suggestions/{id} with status: approved | rejected and optionally an anchor_text override.
4. Apply — confirmed suggestions get woven into the article content.
5. Iterate — new articles automatically trigger a re-analysis of the relevant clusters.

API

Body of PUT /link-suggestions/{id}:

{
  "status": "approved",
  "anchor_text": "Stoßdämpfer wechseln Anleitung"
}

Response of GET /articles/{id}/link-suggestions:

{
  "data": [
    {
      "id": 4711,
      "target_article_id": 88,
      "target_url": "/blog/stossdaempfer-wechseln",
      "anchor_text": "Stoßdämpfer wechseln",
      "source_sentence": "...nach 80.000 km solltest du die Stoßdämpfer wechseln...",
      "confidence": 0.91,
      "status": "pending"
    }
  ]
}

Credits & Limits

• Analyze run: 5 credits per project-wide scan, regardless of the number of articles.
• Approve / Reject: free.
• Async: the analyze job runs >10 seconds and dispatches a queue job — the UI polls progress.
• Rate limit: one active analyze job per project; follow-up requests are queued.
• Inventory minimum: below 5 articles, the semantic analysis returns no useful suggestions — the job returns early with a notice.

Related modules

• AI Content Editor — confirmed links are rendered directly in the editor and are editable.
• Storylines — when you build pillar clusters, internal linking is the natural connection layer on top.
• Content Audit — audit findings often flag "too few internal links" as an issue, fixed here.
• Content Freshness — freshly updated articles often trigger new link suggestions to surrounding clusters.