Audit Search Intent and Cannibalization Before Publishing

• SS

  Surfer SEO

  AI Marketing, SEO & Sales

  SEO content optimization and AI search visibility tracking

  Paid◦Documented
• F

  Frase

  AI Marketing, SEO & Sales

  Content platform for site audits, article drafts, publishing and AI search tracking

  Paid◦Documented

1. 1

   Step 1: Collect a dated result sample for both queries

   Capture the same market, device assumptions and date for each query. Record result URL, page type, title promise and the problem each result appears to solve. Note mixed features such as product pages, videos or local packs because they may signal more than one intent. Avoid declaring intent from wording alone. The sample is ready when every classification can point to visible result evidence and irrelevant results are marked rather than silently discarded.

   Tool: Surfer SEO

2. 2

   Step 2: Classify dominant and secondary intent

   Use the intent-classification prompt separately on each result set. Name the dominant task, secondary tasks, expected page format and stage of awareness. Report disagreement in the evidence instead of averaging it away. Compare the two queries only after each has its own classification. Verification asks whether a reviewer could reproduce the conclusion from the table without relying on your preferred keyword taxonomy.

3. 3

   Step 3: Inventory potentially competing site URLs

   List every existing URL that targets the query family or solves the same reader task. Record its declared purpose, primary headings, canonical target, internal links and available performance evidence. Include URLs that do not rank; cannibalization is about overlapping purpose as well as visible rankings. Do not infer performance when analytics are unavailable. The output is a complete candidate set with unknowns clearly labelled.

   Tool: Frase

4. 4

   Step 4: Build the overlap matrix

   Compare query intent against each page’s purpose, audience, format and unique value. Score overlap descriptively—same, partial or distinct—and add the evidence for that label. Use the query-comparison prompt to distinguish lexical similarity from actual intent similarity. Flag contradictory canonicals or internal anchors, but do not assume they caused a ranking change. The matrix passes review when every action candidate has a traceable scope comparison.

5. 5

   Step 5: Choose the least disruptive supported action

   Use the action-decision prompt to evaluate create, update, merge, redirect and retain. Prefer an update when an existing URL can satisfy the intent without losing a distinct purpose. A redirect requires a destination that genuinely replaces the old page and a preservation plan for valuable material. Document the selected action, rejected options, dependencies and uncertainty. Approval belongs to the site owner because redirects and consolidation can remove discoverable pages.

6. 6

   Step 6: Plan implementation, rollback and measurement

   List copy, canonical, redirect, sitemap and internal-link changes precisely. Capture a baseline before deployment and define observations that would trigger rollback or further review. Use annotations and comparable date ranges where data exists, but avoid promising a ranking increase. Re-crawl the affected URLs after release and verify status, canonical, indexability and links. The decision is complete only when the team can tell what changed and reverse it safely.

• Visit Surfer SEO ↗
• Visit Frase ↗

• Classify intent from real search results · tested on OpenAI GPT-5 (Codex) — editorial dry run
• Compare two queries for overlap · tested on OpenAI GPT-5 (Codex) — editorial dry run
• Decide the correct SEO action · tested on OpenAI GPT-5 (Codex) — editorial dry run

• OutputIllustrative scenario: intent and overlap decision

  Synthetic fixture: two related queries, 12 frozen result rows and three fictional pages. The matrix identified one same-intent page, one partial overlap and one distinct page. The selected action was to update and consolidate into the same-intent page; creating a fourth URL was rejected.

• DatasetTesting limitation

  No live SERP, analytics, Search Console, Frase or Surfer SEO account was used. Results concern only the synthetic fixture run in Codex on 2026-10-03.

✓Last verified Oct 3, 2026