Tracking Organic Traffic Sources and Trends

Analyzing Page-Level Keyword Cannibalization in GA4 for Organic Traffic Decomposition

The shift from Universal Analytics to Google Analytics 4 was never just a migration of dimensions and metrics; it was a fundamental re-architecting of how we are forced to think about user behavior and pathing. For the intermediate web marketer who has already mastered basic traffic filtering and conversion tracking, the real competitive edge in GA4 lies in leveraging its event-driven data model to diagnose one of SEO’s most persistent afflictions: internal keyword cannibalization. You cannot afford to rely on third-party rank trackers alone to spot this. GA4, when queried correctly, offers a ground-truth view of which pages are actually competing against each other for the same search intent, and which one is hemorrhaging authority.

The classic symptom is static or declining organic traffic despite steady or improving keyword rankings. You see positions seven through three for a set of terms, but the aggregate click-through rate feels flat. The culprit is often multiple pages on your site that Google deems equally relevant for a thematic query cluster. To deconstruct this in GA4, you must move beyond the default “Landing Page” report. The key is slicing your organic traffic sources by both landing page and the specific query or user intent pattern. Since GA4 does not surface the search query directly in the way Universal Analytics did (gone are the days of the “Queries” report), you need to combine the “Organic Traffic” source/medium filter with the “Landing Page + Query String” dimension, if you have search console linked.

However, the more powerful approach for the savvy marketer is to use the “Page path and screen class” dimension in conjunction with “Session organic google keyword” (found under the Google Organic Search scope when using Explorations). Create a free-form exploration. Set rows to “Landing page.” Set columns to “Sessions.” Apply a filter for “Session default channel group” containing “Organic Search.” Then, add a secondary dimension of “Page title” or a custom event parameter that hints at the topic. Now examine your data for pages that share an identical or near-identical semantic target. You will see two pages, say `/guides/seo-tips` and `/blog/seo-strategy-2024`, both drawing 300 to 500 organic sessions for similar thematic keywords. The problem is not the volume; it is the distribution. Google is splitting the link equity and topical signals between them.

To diagnose which page is the winner and which is the drain, dive into the “Engagement” metrics within this same exploration. Do not just look at sessions. Look at “Average engagement time per user” and “Key events” (formerly conversions). The page with the higher engagement rate and conversion velocity is the one Google should be indexing as the primary resource. The other page is a parasite on your own domain’s authority. You will also want to check bounce rate trends over a 90-day period. A rising bounce rate on the older page combined with a falling bounce rate on the newer page is a clear signal that the search intent has shifted, and your content architecture has not followed suit.

A less obvious signal within GA4 is “User stickiness” or “Returning user rate” filtered by landing page. A cannibalized page often shows a lower new user percentage because it is siphoning away existing users who would have otherwise landed on a more authoritative page within your cluster. Track the “New user” vs. “Returning user” metric for each suspected cannibal page. If one page has a 70% new user rate and another has an 80% new user rate for the same search theme, the one with the higher new user rate is likely the entry point Google prefers for discovery. The other page is being served to users who have already seen the content elsewhere on your site, which is a poor user experience and signals duplication to the algorithm.

Finally, trend decomposition is critical. Use the “Compare date range” feature in your exploration. Compare the last 30 days to the prior 30 days, then to the same period last year. Look specifically at the trajectory of organic sessions for each competing page. A flattening curve on one page while the other is growing steeply is a sign that the market has consolidated its preference. The correct action is not to delete the weaker page, but to consolidate. You should implement a 301 redirect from the lower-performing cannibal page to the stronger URL, or, if both pages serve slightly different intents (e.g., beginner vs. advanced), you should explicitly cross-link them and disambiguate the title tags and H1s. The goal is to force Google’s index to choose one champion. GA4 gives you the behavioral data to make that choice based on real user economics, not just on opinion. Stop guessing which page is bleeding traffic. Let the engagement data and the source overlap patterns in your exploration tell you exactly where to cut and where to double down.

Image
Knowledgebase

Recent Articles

F.A.Q.

Get answers to your SEO questions.

What role does site search data play in technical SEO audits?
It can uncover indexation and crawlability issues. If users frequently search for content you know exists but returns zero results, it may indicate that your internal search engine isn’t crawling certain pages (like those blocked by robots.txt or with `noindex` tags) or that JavaScript-rendered content isn’t being processed. It also highlights pages with poor keyword targeting that your own site’s algorithm can’t find—a red flag that search engines might struggle too.
What’s the difference between a `noindex` tag and blocking via `robots.txt`?
A `robots.txt` disallow directive blocks crawling but not indexing; if a page has backlinks, Google may still index its URL with a “no snippet.“ A `noindex` tag allows crawling but explicitly instructs search engines to exclude the page from their index. For complete removal, you must first allow crawling with `robots.txt`, then use `noindex` to de-index, then re-block. Misunderstanding this distinction is a common and costly technical SEO error.
What technical elements must be audited to ensure a landing page can be properly crawled and indexed?
Verify the page is not blocked by `robots.txt` or has a `noindex` meta tag. Ensure it returns a 200 OK status code and loads correctly with JavaScript disabled (or that JS is crawlable). Check for proper canonical tags pointing to itself. Validate that internal links to the page use descriptive anchor text and that the page is included in your XML sitemap. Any failure here can prevent indexing, making all other SEO efforts irrelevant.
How do I prioritize which pages to mark up with structured data?
Prioritize based on commercial intent and rich result potential. High-priority targets include product pages, service pages, cornerstone blog content, local business landing pages, and events. Use Google Search Console to identify pages with high impressions but low CTR—these are prime candidates for FAQ or `HowTo` markup to potentially win a rich result. Always start with pages that already rank on page one for valuable keywords to maximize the SERP real estate payoff.
What are common pitfalls in file naming conventions that hurt image SEO?
Avoid generic, non-descriptive names like `IMG_1234.jpg`. These provide zero semantic value. Also, avoid keyword stuffing (`seo-consultant-london-best-seo-expert.jpg`) and using underscores instead of hyphens (Google reads `red_shoes` as one word, `red-shoes` as separate words). The ideal filename is a concise, readable description using target keywords where logical, acting as a secondary relevancy signal for both users and search engines.
Image