The website redesign SEO checklist that prevents a costly reset

Inventory the current site

Crawl the existing public site and export its URLs. Add search landing pages, query data, organic visits, inbound links, and tracked inquiries where available. Mark pages that support sales even when they receive little search traffic, such as a detailed service explanation sent after a call.

Save current titles, descriptions, headings, and main copy. Record the sitemap, robots rules, structured data, and canonical addresses. This inventory provides evidence for deciding what to retain, combine, improve, redirect, or remove.

Give every useful URL a decision

Create a redirect map with one row for every old public address. Keep the address when the subject remains the same. When it changes, send it to the closest relevant new page with a permanent redirect. Do not send every retired page to the homepage because that removes the subject a visitor expected.

Check inbound links before removing a page. If a useful old page becomes part of a broader new guide, update internal links to the new destination and preserve the old address through the redirect. Avoid chains in which one retired URL redirects through another retired URL.

Preserve the search task

A new layout can still fail if the useful answer disappears. Identify the primary search task for each priority page and carry its essential information into the new version. Improve weak or outdated copy, but do not remove relevant sections merely to shorten the design.

Make the service, audience, service area, proof, process, and next action easy to find. Define technical terms. Keep claims supported. When several old pages overlap, create one complete destination and explain each relevant part instead of merging only their titles.

Build search fields into templates

Every public template should support a unique title, description, canonical address, social preview, main heading, and structured data that matches visible copy. Article templates need author and date fields. FAQ schema should be generated from the questions a visitor can read.

Use ordinary anchor links in navigation and related sections. Return useful HTML for the initial response. Keep images sized and described, forms labeled, focus visible, and content usable without motion. These template decisions prevent the same defect from appearing on every page.

Protect measurement and forms

List the events that matter before the old site is replaced. Confirm email clicks, completed forms, bookings, and tool requests. Preserve the landing page, referrer, and campaign fields through the intake flow.

Submit test inquiries to a controlled recipient and inspect the full message. Verify the success page, expected reply text, and any optional booking action. A visual redesign is not complete when its main lead path has not been tested.

Run a release rehearsal

Build the final route list and crawl the preview. Check status codes, titles, descriptions, headings, canonicals, images, internal links, and indexing directives. Test redirect rules against the complete map. Review representative pages on a phone and with a keyboard.

Keep the production domain out of temporary canonical fields. Prevent staging pages from being indexed, but remove that restriction from public production pages. Confirm the sitemap contains only the intended canonical addresses.

Verify after launch

Crawl production as soon as the release is live. Compare the public response with the preview and test every redirect class. Inspect the robots file, sitemap, important canonicals, analytics events, and form delivery again.

Monitor priority pages against the saved baseline. A short fluctuation can happen after a substantial change. Blank HTML, accidental noindex directives, broken redirects, missing links, and failed forms are defects that need action rather than patience.

Example

This example uses illustrative assumptions, not client data. Assume an old site has 36 indexed addresses. The redesign keeps 21 addresses, combines 9 into other pages, and removes 6 pages that have no useful replacement.

• Assumption: 36 old addresses.
• Kept addresses: 21.
• Redirected addresses: 9.
• Intentional removals: 6.
• Decision math: 21 + 9 + 6 = 36 addressed URLs.

The complete total shows that no old address was forgotten in planning. Each redirect and removal still needs an individual relevance check. This is an illustrative migration example, not client data.