What It Does

The AI Linking Strategist analyzes content using topic-based matching to find linking opportunities. When you generate or view content, the system:

  1. Extracts topics from your article
  2. Compares topics against your entire content library
  3. Identifies semantically related pieces
  4. Generates natural anchor text
  5. Validates that anchor text exists in your content

How It Works

Topic Extraction

Every piece of content gets analyzed for key topics. The system stores these topics in the content record for fast matching later.

Example: An article about "Sustainable Swimwear Materials" might extract topics like:

  • Recycled polyester fabrics
  • ECONYL regenerated nylon
  • Plant-based textile alternatives
  • Ocean plastic swimwear

Similarity Calculation

The system uses Jaccard similarity to measure topic overlap between articles. If two pieces share 3 out of 5 topics, that's a 60% match—strong enough to recommend linking.

Minimum threshold: 50% topic similarity. Anything below this gets filtered out.

Anchor Text Generation

The system generates anchor text by:

  1. Finding shared topics between source and target articles
  2. Checking if those topics appear in the source content
  3. Selecting the longest, most descriptive phrase that exists in both
  4. Validating the anchor isn't a generic stop word ("this", "that", "here")

Example: If your article mentions "recycled swimwear fabrics" and you have another article titled "The Complete Guide to Recycled Fabrics," the anchor text becomes "recycled swimwear fabrics."

Validation Filter

Before suggesting a link, the system verifies:

  • Anchor text exists in source article (not fabricated)
  • Anchor text is at least 2 words (prefers multi-word phrases)
  • Anchor text isn't a stop word ("the", "and", "for")
  • Target content is published or ready for linking

Suggestions with invalid anchor text get filtered out entirely—you only see links that can actually be implemented.

Suggestion Types

Outbound Links

What this content should link TO.

When viewing an article, you see suggestions for existing content you should link to from this piece. These appear in the "Linking Suggestions" panel.

Use case: You just published "Summer Swimwear Trends 2026." The system suggests linking to your older articles about specific trend categories mentioned in the new piece.

Inbound Links (Retroactive)

What should link TO this content.

The system identifies existing articles that mention topics covered in your new content. These are opportunities to update old articles with links to new ones.

Use case: Your new article covers "ECONYL fabric benefits." The system finds 5 older articles that mention ECONYL and suggests adding a link to your new deep-dive guide.

Tier requirement: Professional and Enterprise tiers get retroactive linking. Starter tier only gets outbound suggestions.

Package Linking

Links within content packages.

If you generate a content package (pillar + infographics + archetypes), the system automatically suggests cross-links between package pieces.

Package structure:

  • Pillar article (comprehensive guide)
  • Infographic articles (visual breakdowns)
  • Archetype articles (specific use cases)

All pieces in the package get linking suggestions to sibling pieces—but only if valid anchor text exists in the source content.

Resource Linking (Products & Collections)

Links to products, collections, and services.

If you've connected Shopify or WooCommerce, the system suggests links to relevant products/collections based on topic overlap.

Example: An article about "High-Waisted Bikini Styles" gets suggestions to link to your High-Waisted Bikini Collection and individual product pages that match the article's topics.

How it works:

  • Products/collections have topics extracted (from descriptions and tags)
  • System matches product topics to article topics
  • Only suggests products with >20% topic overlap

Where to Find It

Content Detail View

Open any piece of content in the Content Hub and scroll to the "Linking Suggestions" section. You'll see:

  • Outbound suggestions (articles to link TO)
  • Resource suggestions (products/collections to link TO)
  • Package suggestions (sibling content pieces)

Each suggestion shows:

  • Target article title
  • Similarity score (0.0 to 1.0)
  • Suggested anchor text
  • Reason for match (shared topics)
  • Context snippet from target

Implementation

Manual Implementation

  1. View content in Content Hub
  2. Review linking suggestions
  3. Copy suggested anchor text
  4. Find that text in your article
  5. Add link to target URL

Automated Implementation (Coming Soon)

Future versions will support one-click link insertion directly into content HTML.

Best Practices

Anchor Text Quality

Good anchors:

  • "recycled polyester swimwear" (specific, descriptive)
  • "high-waisted bikini styles" (multi-word phrase)
  • "ECONYL regenerated nylon" (branded material name)

Bad anchors:

  • "click here" (generic)
  • "this" (not descriptive)
  • "read more" (overused, no context)

The system filters bad anchors automatically, but if you manually create links, prefer 2-4 word descriptive phrases.

Link Density

Recommended: 3-5 internal links per 1,000 words.

Too few links = missed opportunity for SEO and user navigation. Too many links = looks spammy, dilutes link equity.

Topic Relevance

Only link when semantically relevant. Don't force links to hit a quota. If the system suggests 8 links but only 3 feel natural, use those 3.

SEO benefit: Search engines reward contextually relevant internal links. Random links hurt more than they help.

Package Prioritization

If working with content packages, prioritize package links first:

  1. Link pillar to all archetypes/infographics
  2. Link archetypes back to pillar
  3. Link sibling pieces to each other (if anchor text exists)
  4. Then add regular content library links

Package links create topic clusters—a strong SEO signal.

Tier Differences

| Feature | Starter | Professional | Enterprise |
|---------|---------|--------------|------------|
| Outbound suggestions | ✅ | ✅ | ✅ |
| Resource suggestions | ✅ | ✅ | ✅ |
| Package suggestions | ✅ | ✅ | ✅ |
| Retroactive (inbound) | ❌ | ✅ | ✅ |
| Cross-brand linking | ❌ | ❌ | ✅ |

Starter: New content gets outbound suggestions only. Professional: New content triggers retroactive updates to old content. Enterprise: Multi-tenant linking across brand portfolio.

How Matching Works Under the Hood

Similarity Threshold

  • Minimum: 50% topic overlap required for a suggestion
  • Optimal: 60%+ overlap produces the highest-quality matches

Lower threshold = more suggestions but lower relevance. Higher threshold = fewer suggestions but stronger semantic connection.

Performance

The system analyzes up to 100 candidate articles per query. For large content libraries (1,000+ articles), the system prioritizes:

  1. Recently published content (last 90 days)
  2. High-traffic content (based on your Google Search Console and GA4 data)
  3. Content with the most topic overlap

Older Content Without Topics

If a piece doesn't have extracted topics (e.g., content imported before you started using ILLIXIS), the system falls back to keyword matching based on titles and content text. This ensures your entire library gets linking suggestions, not just newer content.

Automation Schedule

Internal linking runs on automated schedules to keep your link graph current without manual intervention.

Link Graph Rebuild

Schedule: Weekly on Fridays at 3:00 AM UTC

The full internal link graph rebuilds every week. This recalculates:

  • All topic similarity scores between articles
  • Cross-content relationships for your entire library
  • Package link connections

Why Friday: Allows the system to process your week's content before the weekend, so Monday dashboards show fresh linking opportunities.

New Content Link Suggestions

Schedule: Within 1 hour of publish

When you publish new content, the system queues a linking analysis task. Within 60 minutes:

  • Outbound suggestions generate (what to link TO)
  • Inbound suggestions generate (what should link TO this)
  • Resource matches calculate (products/collections)

No action required: Suggestions appear automatically in the Content Detail view.

Broken Link Detection

Schedule: Daily at 4:00 AM UTC

The system crawls all internal links to detect:

  • 404 errors (deleted content)
  • Redirect chains (moved content)
  • Unreachable URLs

Broken links appear in your Content Health dashboard with suggested fixes.

Orphan Page Detection

Schedule: Weekly with link graph rebuild (Fridays at 3:00 AM UTC)

Orphan pages—content with zero internal links pointing to it—get flagged during the weekly rebuild. These represent SEO opportunities since search engines may struggle to discover orphaned content.

Dashboard location: Content Hub → Linking Health → Orphan Pages

Troubleshooting

"No Suggestions Found"

Likely causes:

  1. Content library too small (need 5+ published articles)
  2. New article topics don't overlap with existing content
  3. Extracted topics haven't been generated yet

Fix: Ensure content has been fully processed for topic extraction. Regenerate content if topics are missing.

"Anchor Text Not Found in Content"

The validation filter checks if suggested anchor text exists in source content. If suggestions appear empty, the anchor validation likely failed.

Fix: Edit content to naturally include terms from target article topics. The system will detect the match on next analysis.

"Suggestions Too Generic"

If suggestions link to barely-related content, check topic extraction quality.

Fix: Review the topics extracted for your content. If topics are too generic ("guide", "tips", "best"), regenerate the content with more specific topic guidance.

Future Enhancements

Planned features:

  • One-click link insertion into content HTML
  • Link performance tracking (clicks, engagement)
  • Broken link monitoring
  • Link equity distribution visualization
  • A/B testing for anchor text variations

Related Features

  • Content Packages - Pillar + archetype + infographic structure
  • Content Grading - Checks for sufficient internal linking
  • Site Resources - Product/collection linking for ecommerce sites
  • Preference Learning - System learns which linking suggestions you accept

Ready to lose the stack?

One platform. You approve. ILLIXIS executes. Marketing that just happens.

Join the waitlistNo spam, everUnsubscribe anytime
First 20 founding members: 50% off any plan for your first year.

Marketing, Unstacked.