A website audit that relies solely on PageSpeed Insights and a Screaming Frog crawl misses several critical angles of analysis. An in-depth analysis of a website requires integrating regulatory layers, audience measurement biases, and an interpretation of organic visibility data that has changed with Google’s AI Overviews.
Trackers and consent: the blind spot of any traffic analysis
Before interpreting traffic curves, it is essential to audit the trackers deployed on the site. An audience measurement tracker is exempt from consent only if it is used exclusively to produce anonymous statistics for the publisher. As soon as a cookie also feeds into a third-party tool (remarketing, A/B testing coupled with an advertising platform), the exemption falls away.
The direct consequence: on a site where the consent banner is poorly configured, a significant portion of the actual traffic is never recorded. Page views, bounce rates, and user journeys then rely on a biased sample. We recommend inventorying each tracker, verifying its declared purpose, and cross-referencing discrepancies between server-side data (logs) and analytics tool data.
If you want to frame this step before launching a complete audit, requesting an analysis of my site on On s’appelle allows for an initial diagnosis of the reliability of the collected data.
Digital accessibility and the European Accessibility Act: checking the regulatory scope
Since June 28, 2025, the European Accessibility Act imposes new obligations on certain e-commerce services aimed at consumers. The analysis of a site can no longer ignore this dimension. The first task is to determine whether the activity falls within the scope of the text, and then to examine the relevant journeys (purchase funnel, contact forms, main navigation).
Microenterprises providing services benefit from an exemption provided by the framework. For others, an accessibility audit covers at least the following points:
- Color contrast and typographic readability on key pages (home, product sheet, cart)
- Keyboard navigation and compatibility with screen readers throughout the conversion funnel
- Presence of text alternatives for embedded visual and multimedia content
Integrating this aspect from the audit phase avoids separating regulatory compliance and user experience, two subjects that largely overlap.
SEO audit and organic visibility: beyond raw traffic
Organic traffic alone is no longer sufficient to evaluate visibility in Google. The AI Overviews directly respond to certain informational queries: a site can maintain a stable volume of impressions while recording a drop in clicks. The analysis must therefore separate three metrics in the Search Console: impressions, clicks, and average position, query by query.
An increasing gap between impressions and clicks on informational queries indicates that Google is generating a direct answer. The reaction is not to abandon these keywords, but to identify transactional or navigational queries where the click remains secured, and to focus editorial efforts on those.
The classic technical SEO audit remains relevant: crawling for 404 errors, checking internal linking, analyzing duplicated title and meta description tags, inbound link profile. But the prioritization of corrections must take into account the real impact on clicks, not just on positioning.
Keyword cannibalization and content structure
Two pages targeting the same query neutralize each other in the results. To detect these cases, we cross-reference data from the Search Console (common queries across multiple URLs) with a semantic crawl. Correction involves merging content, a 301 redirect, or a clear differentiation of the intent addressed by each page.
Web performance and user signals: the metrics that matter
The Core Web Vitals remain the foundation of performance analysis. Three indicators structure the diagnosis:
- Largest Contentful Paint (LCP): time to render the largest visible element in the viewport. A slow LCP often indicates unoptimized images or an undersized server
- Interaction to Next Paint (INP): measures responsiveness to user interactions, replacing First Input Delay since March 2024
- Cumulative Layout Shift (CLS): quantifies unexpected visual shifts during loading, often caused by ads or fonts loaded late
A common mistake is to analyze only lab data (Lighthouse, PageSpeed Insights). Field data (Chrome User Experience Report) reflects the actual experience of visitors based on their device, connection, and location. A correct Lighthouse score can mask issues encountered by the majority of users on mobile.
Prioritizing corrections by impact on the journey
Fixing a CLS on an “About” page visited twice a week does not have the same effect as resolving a degraded LCP on the homepage or on a high-traffic landing page. The performance audit benefits from cross-referencing technical metrics with audience data: most visited pages, main entry pages, pages with high exit rates.
A complete website analysis is not a list of scores to improve. It is a diagnosis that articulates the reliability of collected data, regulatory compliance, actual organic visibility, and technical performance prioritized by usage. Each layer conditions the next: without reliable audience data, SEO and UX decisions rely on noise.
