Analyzing the web visibility of a site is no longer just about checking its ranking on a handful of keywords. Since Google launched a dedicated report on performance in its generative AI features (AI Overviews, AI Mode), the very notion of visibility has become fragmented. Measuring the exposure of content in AI modules is radically different from traditional organic tracking, and analysis tools must reflect this reality.
Organic visibility and AI visibility: two metrics not to be confused
The report added to Google Search Console since June 2026 provides only one usable piece of data: impressions in AI features. There are no clicks, click-through rates, or average positions available for these modules. Thus, we measure a raw exposure, without knowing if it generates actual traffic.
Further reading : Discover how to boost your business with innovative digital solutions
In contrast, the traditional organic report from Search Console continues to deliver clicks, impressions, CTR, and average position by query. The two datasets coexist but do not overlap, which requires them to be analyzed separately.
| Metric | Traditional organic report | Generative AI report (since June 2026) |
|---|---|---|
| Impressions | Yes | Yes |
| Clicks | Yes | No |
| CTR | Yes | No |
| Average position | Yes | No |
| Breakdown by query | Yes | No |
This table highlights a major gap: AI visibility remains an indicator of exposure, not performance. A site may appear frequently in AI Overviews without any user clicking through to its pages.
Recommended read : Discover the essential services of the ENT Centrale Marseille to make your daily life easier
To cross-reference this data with more refined traffic and positioning indicators, several SEO platforms offer additional modules. Some, like Virank, centralize the analysis of a domain’s web presence across multiple axes. You can learn more about Virank to evaluate this type of functionality.
Reliability of Search Console data: a documented bias to integrate
Google confirmed in April 2026 that a bug in counting impressions in Search Console affected data over a period covering part of 2025 and early 2026. This logging error over-counted impressions, skewing visibility analyses conducted during this window.
The direct consequence: any organic visibility curve built during this period must be put into perspective. An apparent increase in impressions does not necessarily reflect a real gain in positioning. It may simply result from this technical bias.
Several third-party SEO suites that rely on Search Console data via the API have also been impacted. The historical reports automatically generated by these tools carry the same margin of error. Recalibrating visibility benchmarks prior to the correction becomes a necessary step in any serious audit.
How to identify affected data
- Compare Search Console impressions with crawl data from a third-party tool over the same period to spot abnormal discrepancies
- Isolate spikes in impressions without corresponding variations in clicks or positioning, a probable sign of over-counting
- Consult the official changelog of Search Console, which documents the date of the bug fix and the metrics affected
Web visibility audit: axes that tool lists do not cover
Most comparative guides focus on loading speed, PageSpeed score, or user behavior via heatmaps. These metrics remain useful, but they do not address a fundamental question: is your content truly visible where users are searching?
Web visibility in 2026 plays out across multiple simultaneous layers. Traditional organic (blue links), rich snippets (featured snippets, knowledge panels), AI modules, and social results recently integrated into Search Console form a fragmented ecosystem.
Methodologies for tracking AI visibility: an open project
Semrush, Ahrefs, and tools specialized in GEO (Generative Engine Optimization) have begun to offer modules for tracking visibility in AI responses. The methodologies remain non-standardized and opaque. Two different tools can display contradictory results for the same domain over the same period.
This lack of standard means that an AI visibility audit cannot rely on a single tool. Cross-referencing at least two data sources reduces the risk of erroneous conclusions.
GDPR compliance and analysis tools: a technical selection criterion
The CNIL imposes a strict framework on cookies and trackers used by web analysis tools. Any solution that places a cookie before obtaining user consent exposes the site to legal risk. This criterion eliminates certain default configurations of common analytics platforms.
Alternatives like Matomo, in consent-exempt mode (under specific technical conditions), or server-side analytics tools, allow for collecting traffic data without a blocking cookie banner. The choice of analysis tool directly influences the amount of usable data: a high cookie rejection rate can mask a significant portion of actual traffic.
- Check if the analytics tool used is compatible with the consent-exempt mode defined by the CNIL
- Measure the gap between sessions recorded on the analytics side and server logs to assess the data loss related to consent
- Prefer solutions hosted in Europe to limit data transfers outside the EU
The web visibility of a site is now measured across multiple layers, with tools whose data is neither complete nor always reliable. The AI report from Search Console, the documented impression bug by Google, and the fragmentation of tracking methodologies impose a critical reading of each metric. A credible visibility audit cross-references multiple sources and documents its limitations.