A website redesign is one of the few projects that can quietly destroy years of accumulated SEO value in a single afternoon, and most of the ways it happens are boring: a forgotten redirect map, a staging noindex tag that ships to production, a form that posts to nowhere. The fix is sequence. Here are the 28 steps in the order we run them on every project.
Phase 1: Before anyone designs anything
- Write down why you’re redesigning. ‘It looks dated’ is valid but vague; ‘demo requests flatlined’ is a target. The goal defines every later decision.
- Benchmark current performance: traffic, top landing pages, conversion rates, rankings, Core Web Vitals. Without a before, nobody can prove the redesign worked.
- Crawl the full site (Screaming Frog or similar): every URL, title, meta description. This inventory becomes your redirect map and QA checklist.
- Identify the pages that earn traffic and links. These pages are load-bearing walls: they get preserved or carefully redirected, never casually dropped.
- Interview whoever talks to customers. Sales and support know what visitors don’t understand about the current site better than any analytics view.
- Define the content plan. New copy or migrated? Who writes it, and by when? Content is the #1 cause of redesign delays. Decide before design starts.
- Set the performance budget in the contract: target PageSpeed scores on mobile, in writing.
Phase 2: Design and build
- Design mobile-first: most traffic is mobile and Google ranks the mobile version.
- Build on staging, always. Nothing gets developed on the live site.
- Keep URL structure unless there’s a reason to change it. Every changed URL is redirect work and temporary ranking risk. Change URLs for strategy, not tidiness.
- Build the redirect map as URLs change: old URL to new URL, one row each, no wildcards you haven’t tested.
- Carry over metadata deliberately: titles and descriptions from the crawl inventory, improved where weak, not regenerated from scratch.
- Preserve heading structure on rebuilt pages: one h1, logical h2/h3 order. Builders make it easy to get this wrong.
- Implement schema markup: Organization, LocalBusiness, or product schema as fits, plus breadcrumbs.
- Optimize every image on the way in: WebP, sized to the container, lazy-loaded below the fold. Retrofitting later never happens.
- Rebuild forms and test their destinations: where do submissions go, who gets notified, does the CRM receive them?
Phase 3: Pre-launch QA (the disaster-prevention phase)
- Click every nav item and footer link on staging. All of them.
- Submit every form and confirm the email/CRM arrival, not just the thank-you page.
- Test on real phones, not just the browser’s responsive mode.
- Run PageSpeed on the key templates and hold them against the budget from step 7.
- Verify analytics and conversion tracking fire on staging (in debug mode) so measurement survives the switch.
- Load and verify the redirect map on staging: spot-check the top 50 URLs from your crawl.
- Check the robots meta: staging noindex tags shipping to production is a classic, silent, catastrophic miss.
Phase 4: Launch and the two weeks after
- Launch midweek, in the morning, never Friday. You want business hours and a full team between you and the weekend.
- Immediately after switch: spot-check redirects on production, submit the new sitemap in Search Console, run a fresh crawl for 404s.
- Watch Search Console daily for two weeks: indexing errors, 404 spikes, and Core Web Vitals flags show up here first.
- Compare analytics week-over-week against the benchmark from step 2. Small ranking wobble is normal for a few weeks; cratering traffic means a missed redirect or indexing issue that needs finding today.
- Schedule the 30-day review: conversions vs. benchmark, Core Web Vitals field data, and the punch list of small fixes that surfaced. A redesign ends here, not at launch.
The pattern behind every step: Notice how little of this list is about design. Redesigns fail on process (inventory, redirects, QA, measurement), which is why ‘we know a good designer’ isn’t the same as ‘we’re safe.’ Our redesign projects run this exact sequence, and step 2’s benchmark is how clients hold us accountable afterward.
Planning a redesign?
Send us your current site. We’ll benchmark it, flag the SEO risks specific to your situation, and quote the redesign fixed-price with performance targets in the contract.
