How to fix keyword cannibalisation. The Search Console diagnostic and the consolidation workflow.

What cannibalisation actually is

Keyword cannibalisation is the situation where two or more URLs on the same site compete for the same search intent. Google has to choose which one to surface for a given query. Sometimes it picks one consistently, sometimes it rotates between them week by week, and either way the chosen URL tends to rank lower than a single consolidated page would have.

The clean definition. Two URLs cannibalise each other if they target the same dominant intent on the same primary keyword. Two URLs that share a few overlapping keywords but have genuinely different intents (one informational guide, one commercial service page) are not cannibalising. They are covering related territory.

The mechanism. Google's ranking systems decide one canonical "best" answer per query from a given site. When two URLs both fit the intent, the system splits its confidence between them, which produces a lower effective rank than the same site's strongest single answer would have earned. The two URLs are not competing externally; they are competing internally.

Why it hurts rankings

Three concrete effects we see across client retainers when cannibalisation goes unfixed.

• The link equity gets split. Internal links to the cluster get distributed across two URLs instead of consolidated on one. External backlinks too. The winning URL would have ranked higher with the combined equity than either URL ranks with the split.
• Click-through rate drops on the impressions. When Google does surface one of the two URLs, it sometimes surfaces the weaker one. The snippet and the title that show in the SERP are less aligned with the searcher's question than the stronger URL's would be. Impressions stay; clicks drop.
• The intent signal gets muddy. Google reads two pages on the same intent as a slight signal that the site does not know which one is the right answer. Sites that consistently produce one strong answer per intent get rewarded with cleaner ranking signals over time.

For the broader context on why intent matters more than keyword matching, see understanding search intent. For the clustering work that prevents cannibalisation in the first place, see keyword clustering for SEO. For the topical authority signal cannibalisation interferes with, see topical authority explained.

How to diagnose it in Search Console

The cleanest diagnostic is free. Google Search Console has the data; the workflow takes ten minutes per suspected query.

1. Open the Performance report. Filter the date range to the last three to six months for enough data.
2. Click the query you suspect is being cannibalised. This filters the rest of the report to just that query.
3. Click the Pages tab. Search Console now shows every URL on your site that has picked up impressions for that one query.
4. Read the URL list. If only one URL appears, no cannibalisation. If two or more URLs appear with meaningful impressions (let's say each picking up at least 5 percent of the query's impressions), you likely have cannibalisation.
5. Sanity-check the intent overlap. Open both URLs and verify the dominant intent really is the same. If yes, cannibalisation confirmed. If the pages cover genuinely different intents that share a keyword, leave them alone.

For sites that publish heavily, this manual check does not scale. Tools like Ahrefs (Site Explorer's "Cannibalization" filter), Semrush (Position Tracking's "Cannibalization Report"), or a structured GSC API export into a spreadsheet can flag the candidates at scale. The verification step (intent check on each suspected pair) still has to be manual.

Picking the winning page

Three signals decide which URL keeps the intent.

Signal 1: Current rank

Whichever URL currently ranks higher for the contested query is closer to the SERP. All else equal, the URL that already ranks better is the easier winner to support. Pull the rank from GSC for the last 28 days; if it is rotating, take the average.

Signal 2: Inbound link profile

Whichever URL has more or stronger external backlinks has more underlying equity to draw on. Check in Ahrefs, Semrush or GSC's link report. A URL with three strong external links is worth more than a URL with twenty weak ones.

Signal 3: Conversion data

Whichever URL already converts the traffic that lands on it is the more valuable asset to keep. Pull conversion data from GA4 for the last three months. This signal matters more when the contested query has commercial intent.

Combine the three. The winner is usually obvious: same URL wins on all three signals. When the three signals split (URL A ranks better but URL B has more links and converts better), favour the link profile and the conversion side. Rank can be earned through the consolidation; link equity and conversion track record are harder to rebuild.

The consolidation workflow

Once the winner is picked, the rest is mechanical. Five steps.

1. Audit the losing pages for unique content. Read each losing URL. List the sections, examples, FAQs and entity coverage that the winning URL does not yet contain. This is the salvage list.
2. Merge the salvage into the winning page. Add the unique sections from the losers to the winner. Where there is partial overlap, take the stronger version. The winning URL grows; the losers shrink to nothing.
3. Update internal links to the winner. Find every internal link pointing at a losing URL and change it to point at the winner. This avoids unnecessary redirect hops once the redirects flip.
4. 301-redirect the losing URLs to the winner. Permanent redirect. Not a 302, which signals temporary. See 301 vs 302 redirects for the mechanics. Optionally add a canonical tag on related URLs that should not be redirected but should signal the winner as the preferred version, per canonical tags explained.
5. Monitor and verify. Re-check the GSC query four weeks later. The winning URL should now be the only one showing impressions for the query, and its rank should be at or above the pre-consolidation higher of the two. If not, the consolidation missed an intent overlap or picked the wrong winner.

The whole workflow takes two to four hours per cannibalisation pair, more if the salvage list is long. The four-to-twelve-week recovery window applies the same as for any other consolidation work.

Common mistakes

Perth and WA context

Three patterns from running cannibalisation fixes for Perth and WA clients.

Trade sites cannibalise on service-plus-suburb pairs. A Perth plumbing site that has both /plumber-fremantle/ and /emergency-plumber-fremantle/ where both target the same dominant Fremantle plumbing intent ends up cannibalising itself for the lead query. The fix is usually to merge into one stronger Fremantle plumbing page and let the broader site cover the emergency-specific angle in a sibling. See SEO Fremantle and trades SEO for the live patterns, and keyword clustering for SEO for the prevention.

E-commerce sites cannibalise on category-versus-product pairs. A WA e-commerce store with a category page on "industrial hoses" and a product page that ranks for "industrial hose" (singular) sees the two compete. The fix is usually to reinforce the category page's commercial intent and weaken the product page's broad coverage, so the search system can tell which URL serves which intent. See e-commerce SEO and e-commerce service page for the wider context.

Multi-location sites cannibalise on city-versus-suburb pairs. A regional services business with a Bunbury landing page and a generic South West page can find the two competing for "South West plumbing" type queries. The fix depends on intent: keep the city page for city queries, keep the regional page for region queries, and rewrite both so the difference is clear. See SEO Bunbury, SEO Busselton and multi-location strategy.

For the wider strategic frame, see the Content Strategy pillar. For the prevention work upstream of the fix, see keyword clustering for SEO. For the pruning path that often follows a cannibalisation diagnosis, see content pruning explained. For the canonical signal that complements the 301 redirect, see canonical tags explained.