Overlap and cannibalization diagnosis
Inspect the answers before diagnosing search competition.
Content overlap means that two resources cover some of the same material. Cannibalization is an SEO diagnosis used for pages thought to compete in a way that harms the intended search outcome. Shared titles or changing positions do not establish that harm. Begin with the reader tasks and evidence for the specific URLs.
Free learning from Semantec SEO.
Separate related content from duplicated answers.
Related pages can share entities, introductory definitions and sources while completing different tasks. An overview of a service and a page explaining how to prepare for an appointment may belong together without offering interchangeable answers.
Duplication concerns closely similar material or documents. Intent overlap concerns the reader need being answered. The two can occur together, but neither can be diagnosed from repeated keywords alone. Two pages can use different words while still providing the same full answer.
Harmful competition is a further claim. An editor needs evidence of a problem and a plausible explanation for it, not merely proof that both URLs received impressions. Keep the observation, the proposed cause and the resulting action in separate fields.
Collect a comparable page and query record.
List both URLs, current status, canonical output, title, main answer, audience, evidence, unique material and intended next action. Read the bodies and retained media, not just a sitemap entry. Check if a parameter, alternate route or template is displaying the same document.
For search observations, use an authorized Search Console property and record the date range, search type, query filters, country and device context. Inspect how each candidate appears for relevant queries under comparable filters. Do not claim to have these records when only a public page is available.
Google’s Performance report documentation explains the query and page dimensions and their reporting context. Counts can differ across aggregation choices, and a single result sample cannot represent every search context. Record these limits beside the observation.
Worked example: two similar titles, two possible jobs.
Consider two fictional URLs on a bicycle website: one promises “Bicycle service explained”; the other promises “What happens in a bicycle service?” No live results, customer records or measured performance are supplied for this example. The titles raise a review question but do not answer it.
Read the first page and suppose it defines the service, explains common scope limits and helps a newcomer decide what questions to ask. Read the second and suppose it gives the same definition followed by a detailed preparation and handover process. There is overlap, but the process may provide a distinct answer.
Create a side by side record. Identify the repeated introduction, the unique process details, their supporting sources and the next action. Ask if the first page can complete its introductory task without repeating the entire process. Ask if the second is useful to a direct visitor who already understands the term.
If the process is a supported independent task, retain both, reduce unnecessary repetition and explain the link between them. If the second adds no useful answer beyond the first, propose consolidation. If source or query evidence is missing, record a provisional assessment rather than labeling the pair as harmful competition.
| Observed material | Possible interpretation | Next check |
|---|---|---|
| Similar titles | Possible answer overlap | Read both full answers and compare promised tasks |
| Shared definition | Necessary orientation or avoidable repetition | Check how much context each task needs |
| Distinct supported process | Potential reason to retain separate pages | Check audience, depth and next action |
| No supplied search record | Search effect unknown | Obtain authorized comparable query and page observations |
Read query to page evidence without overstating it.
A query to page record connects a search phrase, the observed destination and the reporting context. It can reveal that a URL appears for queries different from the task the editor intended. It does not directly reveal the private motivation or satisfaction of each searcher.
Consider an explicitly fictional record for 1–30 September 2026, web search, English language queries in Ireland, desktop only: for “bicycle service meaning”, explanation URL A has 120 impressions and process URL B has 80; for “bicycle service handover process”, A has 20 and B has 140. A second fictional period, 1–31 October under the same stated filters, has 90 and 110 for the meaning query, and 30 and 130 for the process query. These invented counts demonstrate the record format only. They are not an export, a forecast or evidence about this site.
The fictional destination mix changes, but the counts do not establish harm. The process page still has a plausible distinct job, the comparison periods have different lengths, and no reader outcome or complete query set is supplied. A reviewer should inspect the answers, qualify the date comparison and seek real authorized evidence before asserting that the URLs damage one another. Record “overlap observed in copy; search competition unproven” rather than calling the title pair cannibalization.
If different URLs appear over time, inspect the pattern across a suitable period and comparable filters. A temporary change can reflect the mix of queries, location, device, season, changed content or other search conditions. A single average position does not isolate the reason.
Check what the reader receives on each destination. Two pages may both appear because they serve useful distinct needs. Also, one URL consistently appearing does not prove the underlying content has no overlap or that a task is completely answered.
Investigate fluctuation as an observation, not a verdict.
Write down what changed: query set, impressions, clicks, destination mix, date or position measure. Then list plausible explanations and the evidence needed to distinguish them. Include site changes, redirects, canonical choices, availability problems and revisions to the offer or topic.
Compare like periods when the business or subject has a seasonal pattern, and mark recent or incomplete data. An unfiltered site wide change should not be attributed to a pair of URLs without narrowing the observation. Search Console is one source of evidence; it is not a controlled experiment.
Avoid promising that merging pages will recover traffic. A consolidation can improve the clarity of a maintained answer while its later search effect remains unknown. Document the intended reader improvement and assess later observations separately.
Choose an action that fits the diagnosed problem.
Retain both pages when each has a supported independent task. Clarify titles, opening scope, cross references and ownership. Revise a page when its promised task and delivered answer differ. Combine only after identifying the material to preserve and the destination that can own it.
Technical duplicates need a canonicalization review. Google’s canonicalization documentation discusses signals for duplicate or very similar URLs. A canonical signal is not a substitute for deciding which distinct reader tasks the site needs.
Do not remove pages merely because a similarity score is high or an export highlights two matching titles. Automated suggestions can identify review candidates; the decision still needs the full answer, source and route context.
Keep an inspectable diagnosis and a safe change plan.
Record the candidate pair, task comparison, repeated and unique material, query evidence and limits, proposed explanation, action, accountable reviewer and expected reader benefit. Use states such as review needed, evidence missing, retain separately or consolidation proposed.
Before changing a live route, check retained source material, current edits, inbound and internal links, media, fragments, navigation, sitemap entries and technical signals. Use a reviewed migration plan for any retired URL. Recheck the public destination and the routes that point to it.
Use topic ownership to choose the maintained answer and split and merge decisions to plan a change. Return to the topical mapping overview or the learning overview.