Discussion starter: a content SEO question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.
- Content SEO
Two URLs keep swapping for the same query. When is that real cannibalisation worth fixing, and when is Google just testing intent?
Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.
1 reply
-
adminAdmin
Swapping on its own is not a problem. It is worth fixing when it costs you something you can measure: the query ranks lower or less consistently than it did when one URL held it, the URL that wins more often is the worse page for that searcher, or Google’s canonical choices show it treats the two pages as near-duplicates. It is more likely intent ambiguity when the results page itself is mixed (different page types from different sites), each of your URLs matches one of those types, and each holds a similar position when it is the one shown.
The qualification is that you cannot observe whether Google is “testing”. You can observe outcomes. So judge the swap by position, clicks and conversions for each URL on the days it appears, and by what the rest of the SERP looks like, not by the fact that the URLs alternate. A swap that persists for months without settling is not really a test. It is a tie, and ties are usually yours to break.
Four patterns that all look like “swapping”
Before deciding anything, work out which of these you are actually looking at. They have different causes and different fixes.
Pattern What the data shows Likely explanation Usually worth fixing? Alternation Daily query and page data shows URL A on some days, URL B on others, rarely both Your pages are competing for one slot and Google keeps changing its pick Depends on intent and cost (see below) Simultaneous listing Both URLs have impressions on the same days, one lower than the other Two slots from your site, or different results for different locations, devices or users Rarely. Two slots is often a win. Reporting artefact Property-level position is steady while page-level rows appear to trade places How Search Console attributes impressions and position, plus noise in low-volume rows No Canonical confusion URL Inspection shows a Google-selected canonical that differs from yours, or the Page indexing report lists “Duplicate, Google chose different canonical than user” Google sees the pages as duplicates and is consolidating them Yes, but it is a duplication problem, not an intent problem The reporting artefact catches a lot of people. Search Console’s documentation on how metrics are counted explains that position is “the topmost position occupied by a link to your property or page in search results, averaged across all queries”. Grouped by property, your site gets one position per search. Grouped by page, each URL is counted separately. A property view can look perfectly stable while the page view looks like a fight.
The mechanism: one slot or two
Two of Google’s documented ranking systems explain most of this. The site diversity system means Google “generally won’t show more than two web page listings from the same site in our top results”. The deduplication systems exist because many matching pages “may be very similar to each other”, and in those cases Google shows “only the most relevant results to avoid unhelpful duplication”.
Put together: two pages that serve clearly different needs can both appear, because each earns its own slot. Two pages that look like the same answer get treated as a duplicate pair, only one can show, and small shifts in signals flip which one wins. That flip is what real cannibalisation looks like in the data.
A hiring analogy helps. Sending two near-identical CVs for one job gets you at most one interview, and the manager may pick a different CV next week. Sending one CV for that job and a different CV for a related job can get you both. Cannibalisation is two CVs for one job. “Google testing intent” is Google not being sure which job the searcher is hiring for, so it keeps interviewing different candidates, including other sites’ pages.
Freshness is a third explanation worth ruling out. Google runs “query deserves freshness” systems for searches where newer content is expected, so an evergreen guide and a recent article can trade places around news or seasonal peaks without either being wrong for the query.
The diagnostic sequence I would run
- Size it. Check impressions and clicks for the query and its close variants over several months. If the whole cluster drives little traffic or revenue, stop here.
- Pull daily query and page data from the Search Console API, because the interface aggregates too much. Classify the pattern: alternating, simultaneous, or both.
- Compare the states. What is the average position and CTR on A-only days, B-only days and both-URL days? If B-only days sit at position 9 while A-only days sit at 4, the swap is costing you. If both-URL days show positions 3 and 7, you have two slots worth keeping.
- Read the SERP. If other sites’ results are uniform (all category pages, or all guides), Google has settled the intent and whichever of your pages does not match is the misfit. If results are mixed, the intent is genuinely split and both of your pages may fit.
- Check duplication signals. Google-selected canonical for both URLs in URL Inspection, statuses such as “Duplicate, Google chose different canonical than user” or “Alternate page with proper canonical tag”, overlap in titles and H1s, and whether internal anchor text for the query points at both pages.
- Check value. Conversion rate and value per organic landing session for each URL. The page that converts better for this query is usually the one you want ranking, provided it matches the intent.
- Check timing. Did the alternation start with a content update, internal linking change, migration or core update? Aleyda Solis’s analysis of the May 2026 core update found shifts driven by intent, market fit and source type. That kind of re-evaluation could plausibly start a swap between two page types, though the analysis does not look at swaps directly.
One reporting caveat: if the URLs trade places around positions 8 to 20, impression counts from before and after September 2025 are not directly comparable, because Google’s removal of the
&num=100parameter changed impression and position reporting for many sites.Deciding which situation you are in
Signal Points to intent ambiguity Points to cannibalisation SERP composition Mixed page types from other sites Uniform page type from other sites Your two pages Different formats or jobs (pricing page and cost guide) Same format, same angle Position when each URL shows Similar for both One consistently weaker, or combined performance worse than when one URL held the query Same-day appearances Common Rare Google-selected canonical Each page is its own canonical One folded into the other, or flipping Conversions Each converts for the visitors it gets The weaker-converting page wins more often Hypothetical example:
A software company’s
/pricingpage and a blog post on “how much CRM software costs” alternate for “crm cost”. Daily data shows alternation with few same-day appearances. On pricing-page days the query averages around position 7 with weak CTR; on blog days it sits around 4. Other sites’ results are mostly independent cost guides, with one vendor pricing page. The reading: Google treats the query as informational and the pricing page is the misfit. The fix is not a merge. Strengthen the guide as the answer, link from it to pricing with descriptive anchor text, and retitle the pricing page around the brand (“[Brand] pricing and plans”) so it stops competing for the generic query.Now suppose the swap is instead between two blog posts, “CRM cost guide” and “How much does a CRM cost?”, written two years apart with the same structure. Same intent, same format, one slot. That is a merge: move anything unique into the stronger post and 301 the weaker one.
Fixes, from least to most drastic
- Leave it and monitor when intent is mixed, both pages hold position and combined clicks are steady.
- Sharpen differentiation. Make each page unmistakably about its own job through titles, H1s and opening paragraphs, and trim the section of the weaker page that duplicates the other’s core answer.
- Re-point internal signals. Send internal anchor text for the query to the preferred URL from navigation, breadcrumbs and related-content modules. Low risk and reversible.
- Merge and 301 when both pages serve the same intent, one is clearly stronger, and the weaker has little unique value or traffic of its own. Expect a settling period.
- Canonicalise near-duplicates that must both exist for users, such as parameter or variant URLs. Remember a canonical is a hint that Google can override.
- Noindex only utility pages that should not rank at all. It is the wrong tool for a page with its own search value.
Check before you merge:
Export every query each URL has earned over at least three months. The “losing” page often brings long-tail queries the winner does not cover, and a merge on a mixed-intent query can remove a second SERP slot. Both can reduce total clicks even though the swap stops.
My decision rule: fix the swap when it costs measurable clicks or conversions and the pages serve the same intent, or when Google’s canonical choices already show it treats them as duplicates. Leave it when the SERP is mixed and both URLs hold position. When the evidence is unclear, start with reversible changes (titles, H1s, internal anchors) and re-read the daily data after a few weeks before reaching for a 301.
Need help with this on your own site?
Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.
Add a reply Cancel reply
Replies are open to approved members. Log in or apply to join.