Keyword Cannibalisation: When Your Own Pages Compete

Most reported keyword cannibalisation is not cannibalisation. Two of your pages appearing for one query is normal, common, and frequently good — Google chooses which page fits each search, and having two relevant pages is not a problem in itself.

Real cannibalisation has a specific signature: two pages that answer the same question, swapping positions over time, with neither ranking as well as one page did or would.

The distinction matters because the standard fix is destructive. Merging or redirecting pages that were not actually competing loses whatever each was earning separately, and it is a hard thing to undo cleanly.

So the diagnosis comes first, and most of the time it ends the investigation.

---

What it actually is

Cannibalisation is when two pages target the same search intent, so Google cannot tell which should rank, and neither accumulates the relevance signals that one page would.

Note the word "intent" rather than "keyword". Two pages can share a keyword and serve different intents entirely — a guide to email deliverability and a page comparing email services both mention the same terms and answer different questions. They are not competing. That is a cluster working correctly, as described in internal linking and topic clusters.

The genuine cases nearly always come from one of four origins:

Origin	What happened
Keyword-list planning	A page was created per keyword variation rather than per question
Forgotten history	Something similar was published three years ago
Scope creep	A page grew to cover the subject of another page
Category and tag archives	An archive page competing with the article on it

The first is the most common by a distance, and it is created deliberately by a research process that treats a keyword export as a content plan — the failure mode keyword research is built to avoid.

---

When two pages ranking is fine

Say this plainly, because a lot of unnecessary merging follows from not knowing it.

It is fine when:

• The intents differ, even if the words overlap — an explanation and a comparison, a guide and a product page
• One page is clearly primary and holds its position while the other appears occasionally, lower
• The query is broad and your pages cover different aspects of it
• Both rank well, which means Google is finding value in both and taking two positions instead of one

That last case is a good outcome, not a problem to fix.

The signal that something is wrong is instability, not co-occurrence. If one page has held the position for months and another shows up at nineteen occasionally, nothing needs doing.

---

The signature of the real thing

Four signs. You want at least two.

Positions swapping. In Search Console, filter to the query and look at pages over time. Page A ranks, then page B, then A again. That flip-flop is the clearest possible signal that Google cannot decide, and it is the diagnostic the whole method rests on.

Neither page ranking well. Both hovering in the teens, where the subject suggests you should do better.

A decline that began at a publication date. A page that ranked, then dropped when something similar went live. Check your own publishing history against the decline.

The wrong page ranking. Google returns your thin secondary article for a query your comprehensive guide should own. This is cannibalisation showing itself as a symptom rather than a pattern, and it is often the first thing anyone notices.

---

Diagnosing it properly

Six steps. Do all six before changing anything.

Step 1 — Confirm the query, not the keyword

Search Console → Performance → filter to the exact query → look at Pages.

If only one of your pages appears, there is no cannibalisation. That ends most investigations, and it takes a minute.

Step 2 — Look at it over time

Compare periods. Three months against the previous three. Is the ranking page changing?

Stability means no problem, whatever the keyword overlap looks like.

Step 3 — Check the intents are actually the same

Open both pages and ask what question each answers.

Different questions, same words, is not cannibalisation. The test is whether one page could satisfy both searchers — the same grouping test used in research, described in search intent.

Step 4 — Search your own site directly

site:yourdomain.com "the exact phrase" shows you every page you have that addresses it, including ones you had forgotten.

Expect surprises here on any site older than a couple of years.

Step 5 — Check the internal linking

Which page do you link to when you mention this subject?

Inconsistent internal linking is frequently the cause rather than a symptom. If half your site points at one page and half at another, you have told Google you are undecided — and the fix is linking, not merging.

This is the cheapest fix available and it is very often sufficient.

Step 6 — Decide whether the outcome is actually bad

Two pages both ranking well is a good result. One page ranking and another appearing occasionally is normal.

Only act if the pattern is unstable, or if the wrong page is winning.

---

The four fixes, in order of risk

Start at the top. Most cases resolve there.

Fix 1 — Change the internal linking (no risk)

Pick the page that should own the query. Then link to it consistently with consistent anchor text from every relevant mention across the site, and remove or repoint the links to the other page.

This is reversible, cheap, and resolves a surprising share of cases, because inconsistent linking is so often the underlying cause.

Give it a few weeks before escalating.

Fix 2 — Differentiate the pages (low risk)

If both pages should exist, make them clearly different.

• Retitle so the distinction is visible in the title
• Narrow one page's scope and remove the overlapping section
• Link them to each other with anchor text that names the difference
• Rewrite the introductions so each states plainly what it covers

Often the pages were always meant to be different and drifted. This restores the intended difference rather than destroying either.

Fix 3 — Merge them (medium risk, irreversible)

When the pages genuinely answer the same question and both should not exist.

The method:

1. Pick the stronger page — usually the one with more links and better historical performance
2. Move the genuinely unique content from the weaker page into it
3. 301 redirect the weaker page to the stronger
4. Update every internal link that pointed at the redirected page
5. Watch for a month, because merges take time to settle

Do not skip step four. Redirect chains and links to redirected URLs are a common cost of sloppy merging.

And be honest at step two. Merging usually means most of the weaker page is discarded, which is the right decision and is why people delay it.

Fix 4 — Remove or noindex (highest risk)

Only for pages with no value at all — a thin variation that was never useful and has no links.

Redirect if there is anything to preserve; noindex if the page has a reason to exist for readers but not for search. Deletion without a redirect throws away whatever the page had accumulated, and should be the last option.

---

Preventing it

Cheaper than fixing it, and it is a planning discipline rather than a technical one.

Plan by question, not by keyword. One page per question, however many phrasings that question has. This is the single most effective prevention and it happens before anything is written.

Search your own site before writing. site: search on the core phrase, every time. Thirty seconds.

Keep a page inventory — URL, the question it answers, its target query. Any site over about thirty pages needs one, and its main value is exactly this.

Link consistently from the day of publication. Decide which page owns which subject, and point at it every time.

Watch scope creep during refreshes. Expanding a page is how it grows into another page's territory — worth checking during refreshing old blog posts.

And handle archives deliberately. Category and tag archives can compete with the articles inside them; decide whether they should be indexed at all, in line with technical SEO.

---

What not to do

Do not merge on suspicion. Merging is irreversible in practice and destroys value if the pages were not competing.

Do not canonicalise between different pages. A canonical says these are the same page. Two genuinely different pages given a canonical to each other is a way of losing one of them.

Do not noindex a page that ranks for anything. Whatever it earns, it stops earning.

Do not treat a keyword tool's cannibalisation report as a diagnosis. These flag keyword overlap, which is not the same as intent overlap and does not consider stability. The flip-flop over time is the diagnosis; overlap is only a prompt to look.

Do not fix five at once. You will not learn which action worked, and one of them will have been unnecessary.

---

Frequently asked questions

What is keyword cannibalisation?
When two pages on your site target the same search intent, so search engines cannot tell which should rank and neither performs as well as one page would. Sharing a keyword is not enough — the intent has to be the same.

Is it bad if two of my pages rank for the same keyword?
Usually not. It is common and often good, particularly if both rank well or one is clearly primary. The problem is instability — pages swapping positions over time with neither ranking well.

How do I check for keyword cannibalisation?
Filter Search Console to the exact query and look at which pages appear, then compare across time periods. If only one page appears, or the same page consistently ranks, there is nothing to fix. Also run a site: search for the phrase.

What is the first thing to fix?
Internal linking. Pick the page that should own the query and link to it consistently with consistent anchor text from everywhere on the site. It is free, reversible, and resolves a surprising share of cases because inconsistent linking is often the cause.

Should I merge pages that compete?
Only after confirming the pattern over time and trying internal linking first. Merging is irreversible in practice — move the unique content, 301 redirect the weaker page, update every internal link that pointed at it, and allow a month to settle.

Can I use canonical tags to fix cannibalisation?
Not between genuinely different pages. A canonical declares two URLs to be the same page, so applying it to two different pages is a way of losing one of them rather than resolving competition.

Do category and tag pages cause cannibalisation?
They can, when an archive competes with the article it lists. Decide deliberately whether archives should be indexed rather than leaving the default in place.

How do I prevent keyword cannibalisation?
Plan content by question rather than by keyword, run a site: search before writing anything new, keep an inventory of what each page answers, and link consistently to the page that owns each subject from the day it is published.

---

What to do next

Pick a query you care about, open Search Console, filter to it, and look at the Pages tab. Then compare the last three months with the previous three.

If the same page ranks in both periods, you do not have a cannibalisation problem on that query — and you have saved yourself a merge you were about to regret.

---

Related guides

• Keyword research that starts with your own data — planning by question rather than keyword
• Search intent: the four types and what they change — the test for whether intents differ
• Long-tail keywords — where over-splitting usually starts
• Internal linking and topic clusters — the structure that prevents it
• Refreshing old blog posts — where scope creep is caught
• Technical SEO: what actually needs fixing — archives, canonicals and indexing
• Why a page is not indexed — a different cause with similar symptoms