Most post-redesign traffic collapses trace back to work that was skipped in the last two weeks before launch. This checklist is the discipline that prevents them — organised by phase, with the non-negotiables marked clearly.
Turn the URL inventory into a decision log
A crawl export is only useful when every important URL has a decision. Add columns for the current page, its target query, organic entrances, enquiries or sales if measured, external links, proposed new URL and the person responsible for approving the change. Mark each row as keep, improve, merge, redirect or retire. This turns a vague redesign plan into a map a developer can implement and an owner can review.
The choice is not always one old URL to one new URL. Several weak guides may serve one intent and belong together on a stronger page. A service page may need to remain even if the new navigation hides it from the first level. For every merge, record what unique information is being carried over and where the old URL will redirect. For every retired URL, check whether internal or external links still point to it.
On this site, an old Leeds agency URL was consolidated into the homepage because both answered the same commercial search. The route list, body links, sitemap and redirect had to agree. That is a smaller example of the coordination a full redesign requires. It does not prove a ranking lift by itself, but it leaves one clear owner for the search intent.
Test the launch as a visitor and as a crawler
Build a short acceptance sheet for the pages that matter most. For each page, check status code, final URL after redirects, title, one H1, canonical, visible main content, primary call to action and structured data that matches the text. Then use a mobile device to try the contact path. A technically indexable page with a form that fails to submit is still a failed launch for the business.
Test the old URLs separately. A permanent move should land on the relevant successor in one hop and preserve the query string where it matters. A redirect that lands on a generic homepage may satisfy a technical check while abandoning the user’s original task. Broken image and script requests also need attention, especially when a cache or CDN makes staging behave differently from production.
After launch, repeat the crawl against the real host with cache disabled and compare the result with the acceptance sheet. Watch Search Console for newly excluded pages and the queries attached to the most valuable URLs. Keep the previous export and redirect map so an unexpected drop can be traced to a specific decision instead of guessed at from an aggregate chart.
Treat a missing redirect map the way you’d treat missing DNS records — the site is not ready to launch.
Before design starts: snapshot everything
Crawl the entire current site and archive it: every URL, title, meta description, heading structure, and word count. Export twelve months of Search Console data — queries, pages, positions — and rank every URL by organic entrances and conversions. This baseline is what "success" will be measured against, and what "we lost something" gets diagnosed with.
Identify your equity pages now: the URLs holding rankings, links, or conversion volume. These get explicit protection plans. A redesign that improves the homepage but breaks three money pages is a net loss nobody notices until the quarterly review.
During design: architecture is an SEO decision
Information architecture determines internal link flow, crawl depth, and topical grouping. Review the proposed sitemap against your keyword clusters: does every commercial intent still have a page? Are equity pages surviving, merging, or dying — and if merging, where does each one’s intent go?
Challenge content reduction early. Designers reasonably want cleaner pages; ranking pages often rank because of content the redesign wants to cut. The compromise is usually structural — tabs, accordions, progressive disclosure — rather than deletion. Content that ranks must survive in crawlable form.
The redirect map is a launch-blocking deliverable
Every old URL needs a destination: itself (unchanged), a 301 to its successor, or a deliberate 410. Map one-to-one wherever possible — redirecting everything to the homepage tells Google those pages’ equity is gone and their relevance signals discarded.
Test the map on staging with a crawler before launch: no chains, no loops, no redirects to 404s. This file is the single highest-leverage document in the entire project. Treat a missing redirect map the way you’d treat missing DNS records.
Launch week: verify, don’t assume
Within hours of launch: crawl the new site fully, confirm robots.txt isn’t blocking (staging noindex tags shipping to production is a classic), submit the new XML sitemap, and spot-check the top 50 equity pages by hand — status codes, titles, canonicals, content presence, schema validity.
Check rendering too: if the new stack leans on client-side JavaScript, verify Google actually sees the content with the URL Inspection tool. A page that renders beautifully for users and empty for Googlebot is a redesign failure with a delay timer.
Post-launch: monitor with rollback triggers
Watch Search Console daily for two weeks: indexing coverage, ranking movement on your equity list, and crawl errors. Some volatility is normal for a few weeks; a 30% visibility drop on money pages is not — it’s a signal to diff those pages against your pre-launch snapshot and fix what changed.
Agree on rollback triggers before launch, not during the argument: which metric drop, sustained how long, forces which remediation. Teams with predefined triggers fix regressions in days; teams without them debate causes for a quarter.
The useful takeaway
Snapshot before, protect equity pages during, treat the redirect map as launch-blocking, verify everything at launch, and monitor against predefined rollback triggers. Redesigns lose traffic through skipped process, not bad luck.
