How do you measure the performance of a semantic cocoon?

In brief: Measuring a semantic cocoon means isolating a fixed URL scope, the central page and the pages built around it, then comparing its visibility metrics in Search Console and its impact metrics in GA4 over comparable periods. Without this stable scope and this before/after comparison, no observed variation can be reliably attributed to the cocoon.

Two tools come into play, and they don't measure the same thing. Search Console shows what Google displays and what visitors click on before arriving on the site. GA4 shows what those visitors do once they're there, right through to conversion. Serious tracking always cross-references these two layers rather than stopping at the first.

Prerequisites before measuring the performance of a semantic cocoon

Before any measurement, five conditions must be met. They determine the reliability of everything that follows, and overlooking them often means comparing data that isn't genuinely comparable.

• Have verified access to the site's Search Console property to view the Performance report, which allows you to analyse data by page and adjust the date range.

• Have access to GA4 with the Google tag correctly installed on all relevant pages: this is a necessary condition for the Landing page report to be populated and usable.

• Establish the exact list of URLs forming the scope to be measured: a central page, its intermediate pages, and the child pages that complete it. This list serves as the reference for every subsequent measurement.

• Define the site's value actions upfront, lead, quote request, purchase, sign-up, which will then be marked as key events in GA4.

• Set a reference date before publication or redesign, to allow for a genuinely comparable period-to-period comparison.

Good to know

Keep these five elements in a single document, updated whenever the scope changes. This document is what will make your comparisons reproducible, month after month.

Step 1: define the cocoon's scope and set the comparison period

In the Search Console Performance report, use the Pages dimension to build a filter that only includes the URLs identified during the prerequisites stage. This filter becomes the measured scope: it, and it alone, should be compared from one period to the next.

Apply exactly the same definition of this scope to every period studied. If a page is added or removed between two measurements, the comparison loses its value: the variation observed may then stem from the scope itself, not from the actual performance of the pages.

Compare comparable periods, before and after a publication or redesign, or one period against the equivalent previous period, rather than an isolated daily variation, which mostly reflects statistical noise. Also bear in mind that the most recent Search Console data may be preliminary: it settles with a slight delay and should not be read as final.

If this set of pages belongs to a broader SEO semantic cocoon architecture, apply this same segmentation logic to each cluster on the site, without ever mixing their respective scopes within a single measurement.

Step 2: track organic visibility metrics in Search Console

Once the scope is fixed, four metrics are enough to establish an initial organic visibility diagnosis.

Clicks, impressions, CTR and average position

These four metrics should first be read at the level of the URL group forming the scope, then at the level of each page and its associated queries, to identify which pages are genuinely driving performance.

Why average position needs to be put into perspective

No: average position is not a fixed ranking for a given query. It is calculated from the highest position of a site link for each recorded impression, then averaged across all those impressions. The same page can therefore appear in a very strong position for one specific search and much lower for a broader variant, without this showing up separately.

This is why it's better to focus on click and impression trends to diagnose a rise or fall, and to use position as a secondary explanatory signal, never as a standalone verdict.

Point of caution

A rising average position doesn't automatically mean more traffic, and a slightly falling position doesn't necessarily mean underperformance. Always cross-reference this figure with click and impression trends before drawing conclusions.

Step 3: monitor the cocoon's visibility in Google's generative results

Since 31 August 2026, Search Generative AI reports in Search Console have been rolled out globally, according to a Google Search Central post published on 3 June 2026 announcing this rollout. They provide a dedicated view of impressions in Google AI Overviews, in Google AI Mode, and in certain generative Discover features.

This view doesn't replace anything: these impressions remain included in the overall performance report as well. It simply isolates what specifically relates to these generative experiences, for a more granular reading when it's useful.

Only use this view if your property has access to it and if the exposure of the scope studied within these experiences is genuinely relevant to the topic at hand: not all cocoons are equally affected by these formats.

The key figure

31 August 2026: the date of the global rollout of Search Generative AI reports in Search Console, according to Google Search Central.

These first two steps provide a picture of visibility, what Google displays and what visitors click on. They still say nothing about what those visitors do once on the site: that's GA4's role, starting from the next step.

Step 4: measure the cocoon's organic entry pages in GA4

Open the GA4 Landing page report and filter or segment the URLs corresponding to the measured scope. This report shows which pages actually receive sessions, across all sources.

Add the Session source / medium dimension to isolate only those sessions whose entry genuinely comes from organic traffic, and exclude sessions from other channels that would distort the reading.

Check that every page within the scope has been visited at least once during the period studied and that it is correctly tagged with the Google tag: this is the condition for the report to be reliable. An untagged page, or one tagged after the fact, will produce incomplete data that understates its actual performance.

Step 5: link the cocoon's organic traffic to business goals

First configure the site's value actions, lead, quote request, purchase, sign-up, as key events in GA4. Without this tagging, none of the following metrics will be available.

Then analyse, for organic sessions and for the pages within the scope, three metrics: the number of key events, the key event rate per session, and the total revenue generated. These three figures answer the question that truly matters: does this traffic drive useful actions, or just visits?

Bear in mind that the credit attributed to a channel can vary depending on the attribution model chosen in GA4: a last-click model and a model distributed across several channels won't tell exactly the same story for the same cocoon.

A purely illustrative example shows this cross-reading: a set of 8 pages goes from 40 to 65 monthly organic sessions over three months, while the key event rate per session rises from 2% to 3.5% on the central page. These figures are not published data: they are used solely to show how to read a rise in traffic and a rise in conversion together, without conflating the two.

The key takeaway

An increase in organic sessions without an increase in the key event rate generally points to a mismatch between content and visitor intent, not a visibility problem.

Step 6: check indexing and internal linking of the cocoon's pages

This last step doesn't measure a result; it checks that the technical conditions for that result are in place.

Checking the indexing of strategic pages

In the Search Console indexing report, check that every important canonical page within the scope is indexed, and investigate the exclusion reasons for any anomalies detected via URL inspection. The goal remains for every strategic page to be indexed, not for every single URL to be.

Don't try to systematically get duplicate URLs or legitimate variants indexed: Google doesn't recommend this, and it needlessly dilutes attention away from the pages that genuinely matter. Indexing makes a page eligible for results; it guarantees neither its display nor its ranking.

Auditing the cocoon's internal linking

Check that every important page receives at least one crawlable internal link, meaning a genuine HTML <a> tag with a populated href attribute, and that the anchor text used usefully describes the destination page. Google uses these links to discover pages and as a relevance signal between them.

This diagnosis assesses the accessibility and context of the scope, but it doesn't in itself prove an improvement in traffic or conversion: it's a technical prerequisite, not proof of performance.

Common mistakes to avoid

Certain mistakes crop up repeatedly and invalidate an otherwise well-intentioned measurement.

• Comparing different URL scopes from one period to the next, which makes any comparison unusable from the outset.

• Judging performance on average position alone, without ever cross-referencing clicks and impressions to understand what's actually happening.

• Treating the most recent Search Console data as final, when it may still be preliminary.

• Measuring organic visibility alone without ever linking that traffic to key events or revenue generated.

• Trying to systematically get every URL indexed, including duplicates or variants that hold no strategic value.

Practical tips for ongoing cocoon monitoring

Tracking that holds up over time relies on a few simple habits, repeated with discipline.

• Keep a reference document listing the scope's URLs and the filters used, so you can reproduce exactly the same measurement each period.

• Schedule a periodic review, monthly or quarterly for example, cross-referencing Search Console and GA4, rather than one-off checks carried out as needed.

• As soon as an overall variation is detected, immediately drill down into page- and query-level analysis to identify the page or pages driving the shift.

• Document every corrective action taken, new internal linking, enriched content, indexing fixes, so you can precisely link an intervention to the effect it subsequently produces.

It's this measurement discipline, more than any tool, that distinguishes genuinely data-driven management from a mere impression of how a set of pages is evolving.

Also worth reading: Topic cluster or semantic silo: what's the difference?.