Core concepts

Workflow

Step 1: Define topic clusters from keyword research

Use keyword clusters from the keyword-research skill as the foundation. Map intent and prioritize pages that serve distinct user problems (informational, transactional, navigational).

Step 2: Create the pillar page structure

A pillar page should:

• Cover the broad topic comprehensively (2,500–5,000 words for competitive topics) and include a clear table-of-contents and short summaries of cluster pages
• Link to each cluster page with descriptive anchor text and contextual placement
• Include short, extractable summary paragraphs (40–120 words) at the start of each subtopic so AI retrieval systems can easily summarize and cite your content (practical adaptation for Web Guide / AI Overviews)
• Target the highest-volume keyword in the cluster and document secondary keywords in the brief

Step 3: Build a content brief template

Add fields that reflect the new search landscape:

• Primary and secondary keywords + intent
• Target length
• Required sections and specific extractable summary for each section (recommended 2–3 sentence summary)
• Structured data required (Article, FAQ, HowTo where relevant)
• Citations & authoritative sources required (list of URLs to cite)
• E-E-A-T notes (author bio, date, data sources)
• Monitoring tags (GSC property, Analytics view, experiment tag)

Example brief additions (extractable summary + citations):

• Extractable summary: "Rate limiting prevents API abuse by limiting requests per key or IP..." (40–80 words)
• Required citation: Link to RFC or authoritative docs for any protocol or standard

Step 4: Implement internal linking (updated)

Rules and additions for 2026:

Run automated audits that check: pillar-to-cluster coverage, orphan pages, and presence of extractable summaries for cluster sections.

Optimizing for AI-organized SERPs (Web Guide & AI Overviews)

Tactics (evidence-based recommendations):

• Provide clear, concise section summaries (40–120 words) and FAQ blocks — AI retrieval prefers short, authoritative snippets (Ahrefs analysis, Mar 2026; Google Web Guide announcement, Jul 24, 2025)
• Use structured data (FAQ, Article) to increase the chance of inclusion in AI summaries
• Keep author and date metadata visible; include citations and links to authoritative sources to strengthen evidence
• Monitor whether pages appear in AI Overviews or Web Guide and A/B test content formats (summary-first vs. narrative-first)
• Prioritize clarity and problem/solution framing — Search Engine Land recommends brand positioning and clarity to be included in AI answers (Apr 2026)

Step 5: Create a content calendar (updated)

Map cluster pages to a production schedule and include explicit measurement & experiment windows for new search surfaces:

• Add a "Web Guide / AI Overview" measurement column to the calendar to record whether a page was surfaced in those features and the date
• Schedule an initial measurement window 14–28 days after publish to record organic rank, clicks, impressions, and AI-surface visibility
• Include monthly editorial QA: accessibility checks (Smashing Magazine guidance), accuracy checks, and human-review for AI-generated drafts

Example schedule (quarter):

• Week 0: Publish pillar page with structured data and extractable summaries
• Week 2: Publish first cluster and begin internal-linking updates
• Week 4: Measure organic performance and appearance in AI features; run content experiments
• Month 2: Refresh underperforming clusters; add citations and expand summaries

Step 6: Content refresh triggers (updated)

Monitor content and refresh when these criteria are met:

Examples

Example 1: Internal linking audit script concept

// Pseudocode for an internal linking audit
type LinkAudit = {
  page: string;
  internalLinksOut: string[];
  internalLinksIn: string[];
  pillarLinked: boolean;
  orphan: boolean;
};

function auditInternalLinks(pages: PageData[]): LinkAudit[] {
  return pages.map((page) => {
    const linksOut = extractInternalLinks(page.body);
    const linksIn = pages.filter((p) =>
      extractInternalLinks(p.body).includes(page.url)
    ).map((p) => p.url);

    return {
      page: page.url,
      internalLinksOut: linksOut,
      internalLinksIn: linksIn,
      pillarLinked: linksOut.some((l) => isPillarPage(l)),
      orphan: linksIn.length === 0,
    };
  });
}

Example 2: Content decay detection

## Content Decay Report — March 2026

| Page | Peak traffic | Current traffic | Decline | Action |
|------|-------------|----------------|---------|--------|
| /blog/api-auth | 3,200/mo | 1,800/mo | -44% | Refresh: update OAuth examples + add extractable summary |
| /blog/jwt-guide | 2,100/mo | 1,900/mo | -10% | Monitor: within normal range |
| /blog/graphql-intro | 4,500/mo | 2,000/mo | -56% | Major refresh or consolidate |

Decision tree

• No content strategy exists → start with keyword research, then build topic clusters
• Keyword clusters ready → map to topic clusters and create pillar page outlines
• Pillar pages exist → create content briefs for cluster pages (include extractable summaries, citations, structured data)
• Cluster pages published → audit and fix internal linking
• Traffic declining → run content decay analysis and refresh triggers
• Too many small pages on similar topics → consolidate into fewer, stronger pages
• New competitor content appearing → analyze their coverage and match or exceed it
• Content calendar is empty → fill with cluster pages ordered by priority score

Editorial workflow & human-in-the-loop (Smashing Magazine alignment)

• Use AI to draft and human editors to verify facts, add unique insights, and ensure accessibility. Smashing Magazine recommends embedding accessibility checks into regular design and editorial workflows (Mar 2026).
• Assign roles: researcher, writer (AI-assisted or human), editor, SEO owner, and measurement owner. Make QA and accessibility part of the publish checklist.
• Monthly review cadence: one full content-audit sprint per month covering accuracy, accessibility, internal links, and AI-surface performance metrics.

Edge cases and gotchas (updated)

1. Pillar page too broad — a pillar that tries to cover an entire industry will be too thin on each subtopic. Keep pillar topics scoped to one product category or problem domain.
2. Over-linking — more than 10–15 internal links on a short page dilutes their value and looks spammy. Limit links to genuinely helpful cross-references.
3. Orphan pages — pages with zero internal links pointing to them are effectively invisible to Google. Run regular audits to find and link orphan pages.
4. Cannibalization — if two cluster pages target the same keyword, they compete with each other. Consolidate or differentiate them clearly. See Moz cannibalization guidance for identification tactics (Moz Blog signal).
5. Stale calendars — a content calendar that is not maintained becomes fiction. Review and update monthly, not just at the start of a quarter.
6. Refresh without re-optimization — updating content without re-checking keyword targeting, internal links, and schema markup wastes the effort. Treat refreshes as a full optimization pass.
7. AI-summarization misalignment — content that reads well for humans may produce poor summaries for AI surfaces. Include explicit, short summaries and citation lines to help AI correctly represent your content (Ahrefs analysis + Google Web Guide announcement).
8. Analytics noise — Search Console reported inflated impressions from May 13, 2025 to early Apr 2026; rely on clicks, CTR, and other stable KPIs while the console stabilizes (Search Engine Land, Apr 3, 2026).

Evaluation criteria

• Every keyword cluster is mapped to a topic cluster with a pillar page
• Pillar pages link to all their cluster pages with descriptive anchor text
• Every cluster page links back to its pillar page
• No orphan pages (every page has at least one internal link pointing to it)
• Content briefs exist for all planned cluster pages and include extractable summaries + citations
• Content calendar is populated for the next quarter and includes an AI-surface measurement column
• Refresh triggers are defined and monitored monthly
• No keyword cannibalization between cluster pages

Versioning

version: 3 updated: 2026-04-06