Keyword cannibalisation is the quiet reason a site with plenty of content and plenty of links still sits on page two. Two or more of your own pages target the same intent, Google cannot decide which one to rank, and the one it picks changes week to week. Rankings oscillate, nothing consolidates, and the page you actually want to rank never gets there.
Indian service sites are unusually prone to it because of how content accumulates: a services page, then a blog post on the same topic, then a city page, then a "why choose us" page that drifted into the same territory. Nobody deleted anything, and now four pages compete.
The 20-minute diagnosis
You do not need a tool subscription for the first pass.
Step 1 — Pull query-to-page data. In Search Console, open Performance, set the date range to the last three months, and export the Queries report. Then export Pages. What you actually want is the pairing, so use the Search Analytics export or the API, or filter query by query for your top 30 commercial terms.
Step 2 — Look for query flip-flopping. For each important keyword, check which URL Google shows. If a single query returns impressions against two or more of your URLs, that is cannibalisation. One URL per query is healthy. Two is a problem. Three means a page needs to die.
Step 3 — Run a site: search for the intent, not the words. site:yourdomain.com plus the commercial phrase tells you how many of your own pages Google considers candidates. If you are surprised by what comes back, that is the finding.
Step 4 — Check the titles. Pages with near-identical title tags are almost always cannibalising. This is the fastest visual scan available: crawl the site, sort by title, and look for clusters.
Deciding which page wins
The instinct is to keep the newest page or the prettiest one. Neither is the right criterion. Keep the page that has:
- The most referring domains
- The clearest commercial intent match
- The URL you would be happy to build links to for the next three years
That is usually the older, uglier page. Links are the hardest thing to replace and the easiest thing to destroy by deleting a URL.
The four fixes, in order of preference
Consolidate. Merge the weaker page's unique content into the winner, then 301 the weaker URL to it. This is the right answer most of the time, and it is the only fix that compounds — one page inherits everything.
Differentiate. If both pages genuinely serve different intent and the overlap is an accident of wording, rewrite one to own its own intent clearly. A page about cost and a page about process can coexist; two pages about "our services" cannot.
De-optimise. Where a blog post is stealing rankings from a commercial page, strip the commercial phrasing out of the post's title, H1 and opening, and link from the post to the commercial page with a clear anchor. The post keeps its informational traffic and stops competing.
Noindex or remove. Reserved for genuinely thin pages with no links and no traffic. Check for backlinks first; a page with referring domains should be redirected, never deleted.
What to do with internal links afterwards
This is the step that gets skipped and it is where half the benefit lives. After consolidation, every internal link that pointed at the retired URL should point at the winner directly — not through the redirect. And every internal link that used the weaker page's anchor is now an opportunity to send a consistent signal to the surviving page.
Pick one primary anchor per target page and use variations of it deliberately. If eleven internal links all say "click here", you have eleven links and no signal.
How often to re-check
Quarterly is enough for a stable site. Any time you publish a blog post that touches a commercial keyword, check the following month whether it started outranking the page it was meant to support — that is the most common way cannibalisation gets reintroduced after a clean-up.
For a full mapping of keywords to the pages that should own them, SEO services in India should start with exactly this audit before any new content is commissioned. See also: the faceted navigation problem nobody fixes before scaling.