Before a migration, analyse the content that ranks

A site migration puts hard-won rankings back in play: every page changes address, often template, sometimes text. It is also the only moment when content can be sorted at no extra cost. You still need to know which pages actually bring visits from Google, and which bring nothing.

This article covers that reading: starting from each page's positions and clicks to decide its fate before the migration. The full audit method, from inventory to action plan, is in our article on the step-by-step content audit; using a third-party tool to enrich it, in analysing your content with Semrush.

The main source is the Google Search Console "Performance" report, filtered by page: the queries that show each URL, its impressions, clicks, click-through rate and average position. Export at least twelve months to smooth out seasonality.

Add:

• the full list of pages, from a crawl of the site (Screaming Frog or similar): a page with no impressions does not appear in Search Console, and that is precisely the one to find;
• conversions per page in Google Analytics 4: a page with few visits that generates quote requests matters more than a widely read page that never converts;
• external links per page (Ahrefs, Majestic, Semrush): a page that receives links is kept, or carefully redirected, even without traffic.

A page on Google's first page for its queries, with steady clicks, has found its form. Carry over its content, title and heading structure without rewriting them; if its URL has to change, its redirect to the exact equivalent is a priority. This is not the time to test new wording on the pages that drive the traffic.

A page that gets many impressions but few clicks, or that sits beyond the first page, has untapped potential. Look at its queries: if they match its topic, the content or title answers the intent poorly and deserves rework during the migration. If they are too broad, or about another topic, the page shows up by accident: refocus its title and introduction on what it really covers, rather than chasing those queries.

When several pages appear for the same query, none of them wins. If each has its own angle, make it visible in the title, the main heading and the links pointing to it. If two pages say the same thing, merge them into the better one and redirect the other. A duplicate content check helps find the clearest cases.

A page with no impressions, no external links and no conversions for a year may be left out: outdated content, old news, a forgotten test page. Remove it, or group its useful content into a nearby page. It then gets a redirect to that page, or a 410 code if nothing replaces it.

This is the most misleading case. A high number of impressions with a distant average position often points to generic queries where the page has no chance: it appears at the end of the results and nobody sees it. Sort its queries by clicks rather than impressions: those that actually bring visits tell you the page's real topic, and that is what the title and introduction should focus on.

Each decision has its counterpart in the redirect plan: a kept page is redirected to its equivalent, a merged page to the page that absorbs it, a removed page to the closest page or answered with a 410. Swan imports Search Console URLs and your URL lists, crawls the old site and the staging site, then suggests for each old page the new page closest in content, which you validate before exporting the redirects.

Next: migrating editorial content and our guide to the redirect plan.