Website analysis: map the pages, explain a variation

Marketing director, SEO lead or executive: before explaining what has moved on your site, you need to know what it is made of. A website analysis starts with a map: group the pages by template and by role, then attach to each group its depth, its internal inbound links, its traffic and its visibility.

That map lets you decide where to fix, at template level rather than URL by URL, and check that the hub, child pages, proof, conversion chain holds. Reading a variation comes next.

What a website analysis brings under control

An SEO-driven analysis measures and explains why a site gains, loses or stalls in organic visibility, then turns those findings into prioritized actions. The goal is not to pile up "warnings", but to identify the points that genuinely cost crawl, clicks or leads. It relies on observable data:

• a crawl for the structure of the site;
• Search Console for impressions, clicks and indexing;
• audience data for engagement and conversions.

The three axes, and what each one gives you to observe

An analysis holds together three axes that interact, and does not treat them in the same way:

Architecture presupposes a map of the site, and that is what is established here. Performance, for its part, is measured (field data, root causes, the trade-off between cost of fixing and expected gain) and belongs to the website performance audit.

Analysis and audit: a question of depth and rhythm

An analysis is an operational reading (site, pages, queries, results) used to decide quickly: what to fix, what to enrich, what to stop. An SEO audit formalizes the diagnosis and the roadmap further; its definition, its triggers and its frequency are set out there. Analysis brings visibility under control day to day; an audit structures a wider plan.

Mapping the site: templates, roles and depth

An actionable map is not limited to "every URL": it groups by template and by role. Without it, the figures cover averages that designate nobody, and the fixes land on whichever URL happens along.

The four page families, and what you attach to each one

Almost any site can be described with four families, each with its own role:

Once the groups are formed, attach four values to each, which make the difference between a list of URLs and a map:

• its depth, in clicks from the home page;
• its internal inbound links, from the navigation and from contextual links;
• its traffic, measured on entry pages;
• its visibility in search results.

A map becomes actionable when it lets you decide "where" to fix, at template level, rather than "which URL" at random. A fix applied to a template propagates to all its pages; the same fix applied to an isolated URL is lost.

The hub, child pages, proof, conversion chain

The map above all serves to make one chain visible: hub → child pages → proof (cases, FAQ, comparisons) → conversion. Reading the chain means checking that each family plays its role:

• the hub directs towards the child pages and distributes authority to them;
• the child pages each cover one sub-intent;
• the proof answers an objection;
• the business page offers an action.

Without that view, you optimize an isolated page when the hub does not distribute authority, or produce new supporting content where the business page already lacks proof. A missing link shows up immediately on the map, while it stays invisible in a page-by-page report.

Depth and distribution of internal links

Depth determines discovery: when a page with something at stake moves away from the home page, discovery slows down, internal linking weakens and the page becomes more "expensive" to crawl. This part of the map is read with three questions:

• Which pages receive the most internal links, and is that consistent with their business value?
• Which strategic pages are too deep, or connected through unintended paths (pagination, filters, internal search)?
• Does internal linking make clear which page is the "reference" when several pages cover a close subject?

The same crossing brings out the orphan pages: those that no internal path reaches from the navigation or from contextual links, even when they appear in a sitemap or are known through campaigns. To spot them, compare the crawl's URL list with the URLs already known:

• the sitemap;
• CMS exports;
• the entry pages measured in audience data.

Those that receive traffic with no internal link identified go to the top of the list. The map names them; attaching them, deciding their fate and the depth benchmark to aim for on pages with something at stake belong to the technical SEO audit, along with the other crawl signals.

The baseline: at constant scope, segment by segment

No explanation of a variation holds without a point of comparison. A useful baseline compares at constant scope (same weekdays, same season, same campaigns) and is cut at least into these segments:

• period A against period B, rather than a month against the previous one, which mixes seasonality with real effect;
• brand against non-brand, because brand often hides structural falls;
• mobile against desktop, because a degradation on mobile can sink conversions without any overall signal moving;
• new against returning, because acquisition content is not built to convert immediately;
• strategic pages against long tail, so as not to drown your business pages in an average.

Document the period and the segments included, and keep the exports: without that, an "improvement" may be no more than a change of scope. Building the baseline, year-on-year comparison and the routine that keeps it alive belong to SEO tracking.

Explaining a variation: impressions, CTR, positions, conversions

A fall is rarely read with a single cause, and almost never at overall level. You start from the strategic pages and their queries, isolate seasonality and on-site changes (first of all an unintentional noindex), then come back up to a family of cause. The map provides the level of reading: a fall attaches to a template, not to the site total. The full reading grid is set out in the method to perform an SEO audit.

Readability and journey: what the user understands, what they abandon

A well-ranked page can still produce nothing. What is measured here is the gap between what the page promises in the SERP and what it delivers once opened. Read template by template, the map brings out three signs of it:

• pogo-sticking (a quick return to the results page), which no tool sees directly and which is approached through weak engagement on organic visits;
• weak entry pages, highly visible but little clicked, or heavily clicked but not engaging;
• a confused journey, with too many steps, no reassurance or forms that are too long.

Diagnosing scannability, friction and the alignment between intent and action belongs to the CRO audit, steered by measured events rather than by taste.

The indicators that steer, and the ones that mislead

An analysis dashboard does not have to be complete: it has to be decidable. Conversely, some very widespread metrics set off more useless projects than any technical anomaly.

The minimum dashboard

Five families of indicators cover what the site does:

Each family is read by group of templates, never as a site total, and attaches to a page family on the map. That is what makes it possible to say which part of the site is progressing, and which one is absorbing the fall.

What the metrics do not prove

These readings come up constantly and often mislead:

None of these reservations forbids acting; each one forbids concluding too quickly. And in B2B, the quality of entries counts more than raw volume.

Governing the decisions that come out of the analysis

Without governance, an analysis produces contradictory optimizations (experience against SEO, content against product) and effects that are impossible to attribute. The minimum viable set-up rests on three elements:

• a single backlog, dated and versioned, gathering every decision that comes out of the analysis;
• measurable acceptance criteria, based on the same sources as the diagnosis, with an observation period fixed in advance;
• a written list of non-priority actions, which avoids scattering and the return of the same subject every quarter.

The critical step remains the trade-off, which is pronounced in three parts:

• what you fix now;
• what you postpone;
• what you will not do, with its written justification.

That trade-off is settled with the impact × effort × risk grid set out in the SEO audit pillar. The analysis stops where the bottleneck becomes structural: when what remains depends on the templates or on the organization, the question is no longer how to read the site but how to change the way it is produced.

Continue reading

Depending on what the map has brought out, three reads take the analysis further:

• Your business pages receive few links: the question becomes that of the links themselves, and internal linking covers hubs, depth and the distribution of authority.
• The fall affects indexing: first check that your pages are discovered; submitting a website covers the discovery, crawl and indexing chain and the ways to secure it.
• Several pages compete for the same intent: the trade-off between creating, optimizing or consolidating a page belongs to the semantic audit.

Obtaining that map grouped by template, with each group's depth, internal inbound links, traffic and visibility, is the task covered by the audit and mapping module.

FAQ: frequently asked questions about website analysis

What is an SEO-driven analysis and what is it for?

An SEO-driven analysis is a measurement and diagnosis exercise that connects architecture, content, technical performance and results. It explains variations in visibility and designates the family of cause before committing resources. It produces a map of the site and a small number of documented decisions, not a list of anomalies.

Which steps should you follow for a complete analysis without an exhaustive technical audit?

A complete analysis without an exhaustive technical audit runs in four stages: map, compare, explain, decide. First group the URLs by template and by role, then establish a segmented baseline at constant scope. Next, go down from the page to its queries, and settle on five to ten decisions, each with a success criterion attached.

How do you produce a reliable, actionable map?

A reliable map groups the URLs by template and by role: hubs, business pages, supporting content, utility pages. Each group is given its depth, its internal inbound links, its traffic and its visibility. It becomes actionable when it designates the template to fix rather than a URL picked at random.

How do you run a UX and SEO assessment without bias from design preferences?

An unbiased UX and SEO assessment relies on behavioural metrics, not on an aesthetic opinion. Track events, scrolling, clicks on calls to action and engagement, segmented by device. Then compare before and after on comparable pages and periods: a page people dislike but that converts is not a problem.

How do you analyse SEO page by page and by search intent?

Analysing SEO page by page means starting from a page, listing its main queries and classifying them by intent: information, comparison, transaction. Then check that the page answers in the format the SERP expects, and cross with audience data to judge traffic quality and conversion. That reading designates the page to handle, not the average.

How do you fix orphan pages without breaking the architecture?

An orphan page is fixed by attaching it to topically close parent pages, not by adding links everywhere. Create links from a hub, a category or a guide, add contextualized sections, then check that a logical path connects it to the home page. If the page has no value, prefer deindexing or a clean removal.