Keyword Cannibalization: Two Pages, One Query, Half the Traffic
Keyword cannibalization is not a ranking penalty. It is a decision you never made. We find every pair of URLs bidding for the same query, choose the page that should own it, and rebuild the internal links and redirects so the winner actually holds position.
- Query-to-URL overlap map built from Search Console, crawl data and backlink profiles
- Merge, split, 301 or re-aim decisions written into a page-by-page brief
- Title, H1 and intro rewrites that separate intent instead of duplicating it
- Redirect chains removed, canonicals corrected, internal anchors re-pointed
- 90-day recheck with before-and-after click and impression windows
No lock-in contracts. One consolidation sprint or an ongoing retainer, scoped after we see the overlap count.
The quiet tax of two pages chasing one query
Most cannibalization is born from good intentions. A product team publishes a buying guide. A content writer adds a blog post on the same topic six months later. The catalog team keeps a category page with an 800-word intro that repeats both. Each page reads fine on its own. Together they give Google three partial answers where it wanted one confident one.
The damage rarely looks dramatic in a dashboard. Average position drifts from 4 to 7. Click-through rate sags because the snippet is now a coin flip between two of your own titles. The page you actually sell from keeps losing the featured snippet to your own blog, so the informational searcher lands on a post with no product, and the commercial searcher lands on a post with no price.
There is a second cost that is easier to miss: authority dilution. Backlinks split. Internal links split. Every new article adds another mouth to feed, and your crawl budget is spent re-indexing near-duplicates. When you later try to rank a single page for that term, it is fighting your own site for the same real estate.
Signals we treat as evidence, not opinion
Rankings that oscillate week to week are the clearest tell: URL A on Tuesday, URL B on Friday, same query, same country. Position stability looks like health but is often just one page winning by accident.
Other dependable signals: two URLs alternating in the same query row of the Performance report, one page ranking for a term it was never written for, an old post outranking a newer commercial page, and a sudden CTR drop on a page whose position did not change.
We confirm with impression share, not vibes. If two URLs both earn impressions for the same query and neither clears a 3% CTR, the pair is a candidate. If one converts and the other only informs, the fix is usually re-aiming, not deletion.
| Keyword | Competing pages | What the data says | Action |
|---|---|---|---|
| stainless steel flange | /flange-guide /products/flanges |
Both sit at positions 6–11 for 7 weeks, combined CTR 1.2%, snippet swaps daily | Merge + 301 |
| custom o-ring sizes | /blog/o-ring-sizes /o-rings |
Blog outranks the catalog page by 5 positions on a commercial query | Re-aim blog |
| export packing | /packing /faq/packing |
Identical H1 and intro; 41% impression overlap, different intents underneath | Split by intent |
| hydraulic hose supplier | /hose /catalog/hose |
Duplicate meta descriptions, no canonical, both indexed | Consolidate |
What you get in the overlap file
Every engagement ends with a document your developers and writers can act on without asking us what we meant. No 90-slide audit, no "improve content quality" advice.
Query-to-URL matrix
A ranked list of every query where two or more URLs earn impressions, with position, CTR, clicks and the date the pair first appeared.
Intent arbitration
Each pair labelled informational, commercial, transactional or navigational, then assigned one owner URL and one supporting URL.
Merge briefs
Section-by-section instructions for folding the weaker page into the stronger one, including what to keep, rewrite, or drop entirely.
Redirect map
One-hop 301s, old-to-new URL pairs, anchor text to update, and the pages that link to the retired URL from the nav or blog.
Re-aimed metadata
New titles, H1s, opening paragraphs and schema for pages that stay, written so they target a distinct modifier instead of the same phrase.
Link equity plan
Where the strongest external links point after consolidation, and which internal anchors should carry the primary term going forward.
The re-aimed metadata work follows the same discipline we apply in SEO Audit Services: one page, one promise, one primary query, with every supporting heading earning its place rather than restating the H1.
How a consolidation sprint runs
Four passes, roughly three to five weeks end to end, depending on how many pairs survive triage and how fast your CMS accepts redirects.
Export 16 months of query data at page level. Join it with crawl output and referring domains so we can see which URL is actually earning links, not just impressions. We also check for faceted navigation duplicates and paginated archives that pretend to be articles.
Score every pair on shared impressions, position gap, conversion value and link strength. Pairs below the threshold are left alone; chasing all of them wastes a quarter. Each survivor gets one owner and one written reason.
Merge, rewrite, redirect. The owner page absorbs the useful sections. The loser vanishes with a single 301, not a chain. Titles and H1s are re-cut so the remaining pages read as a set, not as near-duplicates.
Submit, monitor, compare. We watch the new URL's impressions for 28 days, confirm the redirect resolves in one hop, and re-run the same overlap query to prove the pair is gone rather than merely quieter.
Merge, split, redirect, or re-aim
Four outcomes, four different jobs. Choosing the wrong one is why cleanups fail and the same overlap reappears a quarter later under a new URL.
After the merge: links and redirects that hold
A merge is only finished when the redirect is boring. The most common failure we inherit is a 301 pointing at a URL that itself redirects, so Google sees a chain and discounts both hops. We collapse every chain to a single hop, update the canonical on the surviving page, and remove the retired URL from the sitemap in the same deploy.
Then we fix the anchors. Navigation links, footer links and in-body links that pointed at the retired page must now point at the owner, with the owner's target phrase in the anchor text. If three blog posts all linked to the dead URL with "learn more", they become three different descriptive anchors. That is how a merge turns into a ranking gain rather than a temporary blip.
External backlinks need less work than people assume. A 301 passes the signal, and the strongest referring domains can be re-earned later through outreach or a redirect to a closer match. What matters is that the link lands on a page whose title, H1 and first paragraph all confirm the same topic the linker was writing about.
Finally we write the guardrail: a short intake rule so the next article on that topic links to the owner instead of competing with it. Without that rule, the same cannibalization returns in a quarter with a different filename.
Evidence from recent cleanups
An industrial fastener supplier had 34 cannibalized pairs across a guide hub and a product catalog. We merged 11, re-aimed 19, and left four alone. Organic inquiries moved from 6 to 21 per month over five months.
Industrial supply / 5 monthsA packaging exporter ranked for the same 12 commercial terms with two URLs each. Consolidating halved the indexed page count in that folder; the number of keywords in the top three nearly doubled in four months.
B2B packaging / 4 monthsA SaaS onboarding tool's blog outranked its pricing and feature pages on buying queries. We re-aimed the blog posts to implementation topics and re-pointed 60 internal anchors; clicks from blog to product pages grew 2.4 times.
SaaS / 6 monthsThese are client-reported windows, not projections. We do not promise positions, and we will tell you when a pair should be left alone because both pages serve different audiences.
Scope-based quote, not a package grid
Pricing depends on three things: how many pairs survive triage, whether the CMS lets you redirect without a developer ticket, and whether content has to be rewritten from scratch or merely re-cut. A 40-page site is not the same job as a 4,000-SKU catalog with faceted URLs.
So we quote after the overlap inventory, not before. You get a fixed scope for the sprint, a list of pages touched, and a clear line between what we do and what your team owns — usually CMS edits and final publish approval.
If the overlap is a symptom of a wider crawl or architecture problem, we fold the fix into a full program rather than patching one folder. That is what On-Page SEO Services covers: technical cleanup, on-page work, content clusters and link building under one roadmap, with cannibalization handled as a standing metric instead of a one-off project.
Who this is for
B2B exporters with a guide hub next to a product catalog. DTC brands whose blog outranks their collection pages. SaaS teams that published three versions of the same comparison. Any site where two URLs show up for one query and nobody can say which one is supposed to win.
What we need from you
Search Console access, a crawl or server log if you have one, and one person who can approve redirects. That is enough to start. If analytics is thin, we work from impressions and click data alone — it is noisier but still directionally honest.
Send us the query that has two owners
Tell us the keyword and the two URLs. We will confirm within one working day whether it is a true conflict, and what the cleanest fix looks like.
Questions we answer before signing
How long before a merged page recovers its position?
Usually two to six weeks for the redirect to be processed and the consolidated page to settle. Volatile niches take longer. We compare 28-day windows before and after rather than watching daily rank trackers.
Is cannibalization always a mistake?
No. A brand term appearing on the homepage, a product page and a review page is normal. The problem starts when pages compete for the same non-branded query and neither earns a stable position.
Do you need Search Console access?
Yes, ideally read access at property level. Without it we can approximate overlap from crawl data and third-party visibility tools, but the page-level query data is what makes the diagnosis reliable.
Will you delete pages we still need?
We do not delete anything unilaterally. Every retirement is proposed with its traffic, links and conversion role attached, and you sign off before a redirect goes live.
Is this white hat?
Yes. Consolidation, honest redirects and rewritten metadata are standard publishing hygiene. We do not use doorway pages, cloaking or scaled AI filler to manufacture coverage.
How do you measure success?
Three numbers: the count of queries with two or more competing URLs, clicks to the owner page, and inquiries attributed to organic landing pages. If the first number falls and the other two do not rise, the work is not done.
Do you work monthly or per project?
Either. A one-off consolidation sprint suits sites with a fixed overlap list. Sites that publish weekly usually keep a light monthly retainer so new pairs are caught before they compound.