When the Wrong Page Ranks: A Diagnostic Framework for Competing URLs

A valuable search query can send users to the wrong page. Learn how to distinguish URL competition from intent variation, canonical selection, site architecture and ranking volatility before intervening.

A valuable search query is attracting impressions and clicks, but Google is sending searchers to a page that does not fit the commercial journey. It may be an informative guide when the business needs a product category, or a broad overview when the searcher is ready to enquire. The result can be weaker conversion potential, confusing reporting and uncertainty about what to change.

That does not automatically mean the site has a keyword cannibalisation problem. Google does not operate from a simple one-keyword, one-URL allocation model, and several pages can be valid results for related searches. The page you consider “wrong” may suit another interpretation of the query, or may even perform better commercially than the page you expected to rank.

The useful question is more specific:

Which page should win for this query group, why is another URL being selected, and is the commercial risk large enough to justify intervention?

This article sets out an evidence-led way to answer that question. It separates genuine URL competition from intent variation, canonical selection, internal-architecture signals, template duplication and ordinary ranking volatility.

Define “wrong page” commercially, not technically

A URL is not wrong simply because it is not the one the marketing team had in mind. It is wrong when another page is clearly better suited to the searcher’s likely need and the organisation’s objective, while the selected page creates a material problem.

That problem might be:

• the page does not offer the product, service or next step implied by the query;
• the page attracts organic sessions but has weak landing-page engagement or conversion potential;
• reporting assigns valuable demand to a page that the business cannot optimise effectively;
• two pages appear to serve the same commercial role, creating uncertainty about which should receive content, links and development effort.

For example, imagine a software company with these pages:

• /platform/ — a broad overview of the product;
• /platform/project-management/ — a focused page for teams looking for project-management software;
• /guides/project-management-software/ — an educational guide comparing approaches and features.

For the query group around “project management software”, the second page may be the preferred landing page because it combines clear product relevance with a route to a demo. The guide may still be associated with more recorded visibility, while the platform overview appears intermittently. That is a diagnostic problem. It is not proof that the URLs are splitting ranking signals.

The difference between search-intent drift and page misalignment matters here. A page can be less commercially useful for the business while remaining a reasonable result for a broad or ambiguous query.

The preferred landing-page test

Before diagnosing cannibalisation, test the preferred landing page against three conditions:

1. Searcher and business fit: is the proposed page clearly better for the likely need and the intended commercial journey?
2. Material impact: is the currently selected URL creating a meaningful landing-page, conversion, lead-quality or reporting problem?
3. Plausible mechanism: is there evidence of overlapping page purpose, intent mismatch, canonical or indexation selection, internal-architecture signals or template duplication?

If only the first condition is true, the issue may be a preference rather than a problem. If the first two are true but no mechanism can be identified, more evidence is needed before changing the site. When all three are supported, intervention becomes easier to justify.

Start with query-to-URL evidence

Do not begin by counting how many pages mention a keyword. Define a group of related queries, then examine which URLs are associated with them over time.

The group might include variations such as:

• “project management software”;
• “project management tools for teams”;
• “best project management platform”;
• “project management software demo”.

These queries are related, but they may not express the same need. The final example suggests a stronger commercial action than the broadest version. Brand status, location, device, seasonality and other modifiers can also change the likely intent.

For each query or query group, collect:

• impressions, clicks and click-through rate;
• the URL reported by Search Console;
• ranking history from a consistent source, where available;
• device, location and market;
• landing sessions, conversions, qualified leads, revenue or assisted conversions;
• the page that the business believes should win.

Search Console is useful, but page-level attribution needs careful interpretation. Google may associate performance data with its selected canonical rather than always identifying the literal URL that appeared or received the click. The relationship between Search Console URL attribution and canonical selection is therefore an important part of the diagnosis.

Use URL Inspection, analytics landing-page data and independent rank checks to test the pattern. None provides a complete historical record of every URL shown for every impression. A URL that appears to “own” a query in one report may not be the literal result shown in every search.

A worked diagnostic example

Suppose the software company reviews six months of data for the query group above.

• Months 1–2: /guides/project-management-software/ receives most recorded impressions and ranks between positions 6 and 10.
• Months 3–4: /platform/project-management/ appears more often and reaches positions 4 to 7 for the more commercial variants.
• Months 5–6: the guide returns for broad queries, while the product page ranks for “project management software demo” and “project management tools for teams”.

The first conclusion should not be “Google is confused”. The pattern may show that the pages serve different interpretations of the query group. The guide may be appropriate for research-oriented searches, while the product page is better for users evaluating a platform.

The marketing team should then ask:

• Does the guide have a distinct research purpose, or is it a lightly rewritten version of the product page?
• Does the product page clearly explain the use case and next step?
• Do the URLs alternate for the same specific query, or only across queries with different modifiers?
• Which page produces stronger demo starts, qualified leads or assisted conversions?
• Did a navigation, template, canonical or content change coincide with the URL change?

If the guide ranks for broad research queries and the product page ranks for high-intent variants, the variation may be healthy. If both URLs alternate for the same high-value query and the guide has a materially weaker conversion path, there is a stronger case for investigation.

Test the main hypotheses

1. Genuine overlap in page purpose

This is the closest match to what practitioners often call cannibalisation. Two pages may target the same underlying need, use substantially overlapping language and offer no clear reason to exist separately.

Look for:

• substantial overlap in the query groups each page attracts;
• similar titles, headings, introductions and calls to action;
• the same commercial proposition presented through different URL paths;
• alternating rankings for the same query variants;
• no distinct audience, stage or task that justifies keeping both pages.

These findings support an overlap hypothesis, but they do not prove that one page caused the other to lose visibility. Relevance, content quality, freshness, external references and technical state may also differ. Treat the overlap as a reason to investigate, not as a diagnosis on its own.

2. Search-intent mismatch or change

A broad query can support several legitimate results. Search engines may show a mixture of product pages, guides, comparisons, local results or support content because the wording does not reveal one clear need.

Compare the pages with the search results themselves. Are the results consistently commercial, or do they contain several page types? Has the result mix changed over time? Does the selected URL answer a different but reasonable interpretation of the query?

A page that is less suitable for the intended commercial journey may still be a valid result for another searcher. Forcing one URL to rank for every related phrase could remove useful coverage rather than improve it.

3. Canonical or indexation selection

A canonical declaration tells search engines which URL a site considers the representative version of duplicate or substantially similar pages. Google may choose a different canonical from the one declared.

Check:

• the canonical declared on each relevant page;
• the canonical selected by Google in URL Inspection;
• whether the pages are genuinely duplicate or substantially similar;
• whether the preferred page is indexable, crawlable and available with a successful status code;
• whether redirects, noindex directives or parameter variations alter the URL set.

A canonical mismatch is useful evidence, but it is not a complete explanation of ranking behaviour. The selected canonical is not necessarily the URL displayed in search results, and canonicalisation is primarily a duplicate-handling mechanism. It should not be used to force genuinely different page purposes into one representative URL.

Do not use robots.txt as a canonicalisation method. It can prevent crawling without telling Google which accessible page should represent a duplicate set. Likewise, noindex is not the preferred way to nominate a canonical within a duplicate group.

4. Internal-architecture or linking bias

Internal links help search engines discover pages and provide context about linked destinations. A site that repeatedly links to one URL with commercially important anchor text may be signalling a different preferred destination from the one the business wants to rank.

Review:

• links from the main navigation, category pages and high-authority content;
• anchor text and surrounding copy;
• links from pages already attracting relevant organic demand;
• orphaned or weakly connected preferred pages;
• template modules that repeatedly promote one URL over another.

Link counts alone do not prove causation. A page may have more internal links because it is already considered more important, while content quality, external links, freshness and page purpose also influence its performance. The useful finding is not that one URL has more links than another. It is whether the site’s architecture consistently communicates a different preferred destination from the one the business wants to rank.

5. Template duplication

Large sites often generate ambiguity through templates. Many pages may share the same primary copy, metadata, navigation modules and calls to action, with only a location, product or category label changed.

That does not make every templated page harmful. The question is whether the differences are substantial enough to support distinct search demand and distinct user needs. Compare the pages at template level as well as individually:

• what content changes between URLs;
• whether the changed content answers a different question;
• whether page titles and headings distinguish the purpose;
• whether internal links reinforce meaningful categories or merely generate URL variations.

A template problem may require a structural fix. If only a small number of pages have a material mismatch, page-level clarification may be safer than redesigning the entire system.

Decide whether the issue is material

Not every URL change deserves engineering time. Prioritise the issue by combining demand, commercial value and confidence in the diagnosis.

A high-priority case usually has several of these characteristics:

• the query group has meaningful impressions, clicks or commercial value;
• the selected URL is clearly weaker for the intended journey;
• the preferred page is indexable and capable of satisfying the query;
• the mismatch is sustained rather than a short-lived ranking fluctuation;
• there is evidence of a plausible mechanism;
• the current page produces weaker conversions or lower-quality leads.

Keep the conversion test open. The supposedly wrong page may convert better because it sets expectations more accurately, answers an earlier question or attracts a more qualified audience. Organic landing-page conversion data can also be incomplete when users return through another page, convert across sessions or are measured through assisted conversions.

Choose a proportionate intervention

The evidence should determine the action.

• Clarify page purpose when the pages are both useful but their roles are blurred. Give each page a distinct audience, question and next step.
• Improve the preferred page when it is the right destination but does not yet satisfy the query as clearly as the competing URL.
• Revise internal links when navigation and contextual links consistently favour the wrong destination. Change the architecture deliberately rather than simply adding more links.
• Correct canonical or indexation signals when duplicate or near-duplicate URLs are being handled inconsistently and one representative URL is genuinely appropriate.
• Consolidate or redirect redundant pages only when they have no distinct purpose and the business case supports removing the alternative.

This is different from applying a standard merge-or-redirect recipe to every instance of URL variation. The broader content-consolidation decision should follow evidence about page purpose, demand and risk, not just overlapping keywords.

Implement and validate safely

SEO changes become difficult to interpret when several variables move at once. Use a controlled sequence:

1. Establish a baseline. Record query-level URLs, rankings, impressions, clicks, landing sessions and commercial outcomes. Capture the relevant canonicals, indexation state and internal-link pattern.
2. State the hypothesis. For example: “The guide ranks for the demo query because the product page is weakly linked and the two pages have overlapping copy.” This is a working explanation, not an established fact.
3. Make one coherent change. Improve the preferred page, adjust the relevant architecture or correct the technical signal. Avoid combining an unrelated rewrite, URL migration and navigation redesign unless the implementation requires one coordinated release.
4. Complete technical QA. Check status codes, redirects, canonicals, indexability, internal links, XML sitemaps, analytics landing pages and conversion tracking.
5. Monitor the original query group. Track which URL appears, not only aggregate clicks. Compare visibility, landing-page fit and conversions with the baseline.
6. Allow for confounding factors. Competitor activity, seasonality, algorithm updates, freshness and SERP changes can affect the result. A short-term URL switch is not proof that the intervention worked.

A stronger causal diagnosis requires a sustained, query-specific change that follows the intervention, while other explanations are considered. Even then, live-site SEO rarely offers perfect experimental control.

Limitations to keep in view

• Volatile SERPs: short-term URL changes may reflect normal ranking variation, freshness, competitors or algorithm updates.
• Mixed-intent queries: broad wording can legitimately produce several page types.
• Branded searches: a homepage, support page, product page or location page may be the appropriate result even when the business prefers a commercial landing page.
• Different user needs: two pages can serve complementary stages of a journey. Consolidating them may remove useful coverage.
• Attribution limits: Search Console, rank trackers and analytics do not provide a perfect historical record of every literal URL shown or every assisted conversion.
• Unexpected commercial performance: the page that appears less suitable may produce better-qualified leads or stronger assisted conversions.

These limitations do not make diagnosis impossible. They mean that URL competition should be treated as a testable business hypothesis rather than an automatic technical failure.

What a sound diagnosis looks like

The most useful conclusion is not “three URLs rank for this keyword”. It is something more specific:

For this query group, the product page is the better landing page because it matches the searcher’s commercial need and produces stronger qualified actions. The guide has ranked for the same high-value variants for four months, shares its primary purpose and copy with the product page, and receives stronger internal prominence. The evidence supports clarifying the product page, revising the relevant links and monitoring URL selection before considering consolidation.

That conclusion identifies the preferred page, the commercial risk, the likely mechanism and a proportionate next step. It also leaves room for evidence to disprove the initial assumption.

Liquid Silver approaches this work by connecting query data, ranking history, page purpose, internal architecture, canonical and indexation signals with landing-page and conversion evidence. Specialist support is particularly useful when a site has a large URL set, several plausible causes, weak historical data or fixes that span content, technical SEO, analytics and development teams. The objective is not to make every query resolve to one URL. It is to establish which page should win, make that decision safely and validate whether it improved the business outcome.