Website Content Audit: Prioritise Pages With Search Console, GA4 and AI

Build a URL inventory that includes pages with no recorded traffic

Start with a CMS export or crawl containing the pages you intend to assess. Add the full URL, page type, intended reader, purpose, owner and last substantive review date. Where available, include HTTP status, declared canonical and indexability evidence from the crawl.

Compare that inventory with your sitemap and reported URLs. A URL missing from Search Console or GA4 has no observed row in that response. It does not automatically have zero traffic, and it might be new, excluded, untracked or outside a limited result set.

SourcetoMCP’s Search Console operations can read search performance, sitemap records and reported URL inspection results. They do not provide a complete website crawler or inspect every page’s prose. Supply the current page text or review it in the browser for factual and editorial checks.

Google’s helpful-content questions are useful for reviewing whether a page has an original, complete answer for its readers. Assign a specific missing fact or example rather than an unsupported “quality score”.

Collect comparable performance evidence from each source

Choose two complete periods and record the property, country, device and search type. Use longer or year-on-year comparisons when seasonality and the available data support them. Keep recent incomplete days out of both periods.

Retrieve Search Console page clicks, impressions, CTR and average position. Investigate selected pages with query-and-page reports afterwards. The Search Analytics API returns top rows and does not guarantee a complete inventory. The connector caps each request at 1,000 rows, so retain warnings and narrow the follow-up scope.

In GA4, use landingPage and the chosen session source or channel dimensions when the question concerns entry-page outcomes. Apply supported dimension filters before the row limit. A pagePath report describes pages encountered during visits and can allocate sessions across multiple page rows, so it needs a different interpretation.

Record the key-event configuration behind any outcome measure. In this connector, the conversions metric is an aggregate alias for GA4 key events. It does not establish how many qualified leads or paid customers each page caused.

Match URLs without collapsing different pages

Search Console supplies full page URLs, while Analytics landing-page values can use paths and query strings. Match the host and path deliberately. Retain the original values beside the matching key so someone can inspect what was combined.

Document redirects, trailing slashes, case and query-parameter rules. A campaign parameter can be irrelevant to the content inventory, but a product or language parameter can identify a different page. Removing all parameters blindly can merge distinct content.

Google’s canonicalisation guidance explains the representative URL used for duplicate content. A declared canonical, Google-selected canonical and GA4 landing path can differ. Keep those observations separate until you establish why.

Use the Search Console versus GA4 guide when clicks and sessions differ. Their totals measure different activity, so a mismatch alone does not identify a broken page.

Prioritise specific page decisions with business context

For pages losing search clicks, use the content-decay investigation to compare complete periods and separate stale answers from seasonal demand before proposing a refresh.

The following hypothetical audit covers three pages over one complete 28-day period. Key events use the same documented property definition. The examples illustrate prioritisation, with no claim that they represent customer results.

The legacy page has the least search traffic but still records meaningful activity across channels. Removing it based only on four search clicks would ignore the rest of the evidence. Check links, customer use and other acquisition sources first.

For the setup guide, a planning scenario of 3% CTR at unchanged impressions gives 240 clicks, or 80 more than the current 160. That is an arithmetic scenario. It does not show that correcting the instructions will produce the assumed CTR.

Assign the update to an owner and estimate effort. Keep a factual correction ahead of a speculative traffic rewrite when the old instructions send readers through the wrong setup.

Ask AI for a reviewable audit and an explicit evidence gap

Audit the supplied URL inventory for [site]. Make no website or file changes. Compare [complete periods] using Search Console web search with [market filters] and GA4 organic landing-page results with the documented key-event definition. Keep source URLs, matching rules and response warnings. Use the supplied content and crawl evidence to propose keep, refresh, merge or further investigation. Give each action its URL, observed evidence, unresolved checks, owner and effort. Mark missing data as unavailable. Do not infer query-level conversions or recommend automatic deletion.

Save the approved queue in a workbook or report with the inventory version and retrieval date. The SEO report template provides a structure for evidence and actions. Updating a connected reporting file is a separate approved step from gathering the figures.

For a small set of existing pages ready for edits, use the SEO quick-wins workflow. Keep this wider audit responsible for inventory, factual accuracy and pages that do not yet appear in the performance sample.

Review merges and measure approved changes

If two pages appear to compete, inspect their questions, audiences and content before merging them. The keyword-cannibalisation guide covers this investigation. Shared queries alone do not establish that two pages serve the same purpose.

A merge needs an agreed destination, a plan for useful content and links, and appropriate redirect or canonical handling. Google’s redirect documentation explains permanent moves. An audit recommendation does not itself implement those changes through SourcetoMCP.

Record the approved edit, old and new URLs where relevant, publication time and follow-up period. Compare search visibility, relevant visits and outcomes using the same definitions, and preserve an inconclusive result when the evidence is too small to support a conclusion.