A B2B site with 180 live URLs launches with a cleaner design, 47 fewer pages and a rewritten service hierarchy. Three weeks later, organic clicks are down 32% and demo requests have fallen from 18 to 10.
The new site may look better. It is still a commercial failure.
A website redesign without losing SEO is a preservation job before it is an optimisation job. Record what currently earns visibility and leads, decide where every useful page and search intent will live, redirect precisely, then test the evidence rather than trusting the build.
Why website redesigns lose SEO
Google does not penalise a company for changing its colour palette or page layout. Losses usually come from changing several search signals at once without controlling them.
A ranking service page gets a new URL. Its detailed copy becomes three vague paragraphs. Internal links disappear with the old navigation. The canonical points elsewhere. Then the old URL redirects to the homepage because nobody agreed on a proper destination.
Each decision looks small inside a design project. Together, they tell search engines that the useful page no longer exists.
The safest position is simple: do not change a URL unless there is a clear reason. A shorter slug is rarely worth putting an established page through reprocessing. When the URL must change, preserve the page's purpose, topical coverage and internal importance at its new address.
This is also why SEO preservation and conversion improvement need separate acceptance criteria. A page can retain its rankings and still convert badly; it can also convert beautifully after losing half its qualified traffic. Use a conversion-focused redesign process to improve the commercial journey, but do not let those changes erase proven search demand.
Most losses are predictable. The redesign mistakes that quietly cost leads usually start with missing evidence, unclear ownership or an incomplete URL map—not a mysterious algorithm update.
Audit before the first wireframe
SEO cannot be preserved from memory. Build a baseline while the current site is still crawlable and its analytics are intact.
Pull at least 12 months of data so seasonality does not distort the comparison. Combine five sources:
- A full crawl for URLs, status codes, titles, headings, canonicals, indexability and internal links
- Google Search Console for clicks, impressions, queries, landing pages and indexing
- Analytics for organic entrances, enquiries and assisted journeys
- The CMS and XML sitemap for pages a crawler may not discover
- Backlink data and, where available, server logs for externally linked and actively crawled URLs
Do not treat traffic as the only measure of value. A technical article with 4,000 visits and no commercial contribution may deserve less protection than a niche service page with 300 visits and six sales-qualified leads. A practical B2B conversion audit helps separate busy pages from pages that move buyers forward.
The output should be one inventory, not five exports. Give every URL an owner and record:
- Organic clicks and qualified conversions
- Important queries and current search intent
- Referring domains and useful internal links
- Content type, template and canonical URL
- Proposed action: keep, improve, merge, split or retire
- Proposed destination and redirect requirement
- Risk level and named approver
The gap between a light reskin and the builds in these selected website projects is substantial. Define which templates, CMS collections and paths are changing before anyone estimates the preservation work.
What one lost page can cost
Suppose a specialist B2B service page receives 600 organic visits a month and converts 2.5% into enquiries. That is 15 enquiries.
If 20% become qualified opportunities, 25% of those close, and average first-year revenue is £30,000, its expected monthly revenue contribution is:
15 enquiries × 20% qualified × 25% won × £30,000 = £22,500
Now assume the redesign cuts that page's organic traffic by 40% while its conversion rate stays flat. It produces nine enquiries and an expected £13,500. The avoidable difference is £9,000 a month.
That is not a forecast to book as revenue. It is a risk value that tells the project team where extra review, content parity and redirect testing are justified.
Map every URL and its search intent
The URL map is the control document for the redesign. Every existing indexable URL needs a decision, including PDFs, campaign pages and orphan pages found in Search Console or server logs.
Use five actions:
| Action | Use it when | SEO treatment |
|---|---|---|
| Keep | The page and its purpose remain useful | Retain the URL and improve carefully |
| Merge | Two or more pages satisfy the same intent | Build one complete destination and redirect each source |
| Split | One page serves several distinct intents poorly | Create focused pages; redirect the old URL to the closest primary successor |
| Retire | The content has no demand, links or valid replacement | Return 404 or 410 and remove internal links |
| Replace | The same intent moves to a new URL or template | Preserve useful content and apply a direct permanent redirect |
Mapping by subject alone is not enough. “Cyber security services”, “cyber security consultancy” and “cyber security audit” may sound related, but their searchers can expect different services. Redirecting all three pages to a generic security page can turn valid redirects into irrelevant destinations.
For each important old page, compare the proposed replacement side by side. Check the primary query group, intent, core claims, proof, FAQs, downloadable assets and internal links. The new page does not need the same word count. It does need to remain a credible answer to the demand the old page met.
This is where redesign teams often over-edit. Strong copy becomes generic because it must fit a component. Detailed case evidence is cut because the new layout has less room. A new tone can be agreed through a working brand personality framework, but voice work should not strip out the terminology and specificity buyers search for.
Freeze the approved map before development finishes. Any later URL change should require an owner, a reason and an updated redirect rule. A shared spreadsheet with 100% coverage is more useful than a polished strategy deck with gaps.
The redirect plan for a website redesign without losing SEO
A redirect is a routing decision, not a cure for deleted relevance.
Use server-side 301 or 308 redirects for permanent moves. Each old URL should go directly to the most relevant live replacement in one hop. Do not create old-to-intermediate-to-final chains, and do not redirect every retired page to the homepage. Google’s guidance for URL-changing site moves warns that irrelevant bulk redirects may be treated as soft 404s.
Test more than the neat URLs in the spreadsheet. Include trailing-slash variants, upper-case paths, legacy HTTP versions, common parameter versions and old PDFs with external links. Check that redirects preserve query strings only where those parameters still have a purpose.
Then remove the need for search engines and users to pass through redirects internally:
- Update navigation, body links, breadcrumbs and XML sitemaps to final URLs
- Give each new page a self-referencing canonical
- Keep only canonical, indexable 200-status URLs in the new sitemap
- Update structured data, hreflang and image references where relevant
- Remove links to retired pages
Keep the redirect rules for at least one year, as Google recommends; longer is sensible when old links, bookmarks or sales documents remain in circulation. Maintain the map beside the code so a later platform update does not quietly remove it.
QA the redesign against SEO evidence
SEO QA needs pass-or-fail criteria. “The pages look fine” is not one.
Before release, crawl the staging site and compare it with the approved inventory. Staging should be blocked from search results, but the QA team must be able to crawl it. Check rendered content, not only what appears in the CMS.
For priority pages, verify:
- The intended URL returns 200 and the old URL resolves to it directly
- The title, main heading and body still satisfy the mapped intent
- Canonical, robots directives and sitemap URL agree
- Navigation and contextual links point to the final URL
- Structured data matches visible content
- Mobile rendering exposes the same substantive content
- Analytics and qualified-lead events fire once, with the right source data
A sensible go/no-go standard is zero redirect loops, zero priority pages with accidental noindex directives, zero sitemap URLs returning anything other than an indexable canonical 200, and zero internal links to mapped legacy URLs. For high-risk pages, 100% of redirects should be tested—not sampled.
If an external team is handling the website design and redesign work, put those criteria in the brief and name the person who can stop release. Responsibility split between a developer, SEO consultant and marketing lead is how critical checks become “someone else's job”.
After release, recrawl the old URL set and the new site. Submit the clean sitemap in Search Console, inspect representative high-value URLs and watch organic landing-page performance, indexing, 404s and server errors. Review daily during the first week, then weekly until performance settles.
Measure against the baseline by page group, not only site-wide. Branded demand can hide a 50% fall across non-brand service pages. Compare clicks, impressions, average position and qualified conversions for retained, merged and replaced pages separately.
Some fluctuation is normal while search engines recrawl and reassess changed pages. A continuing fall concentrated in one mapped group is not “just the algorithm”. Check redirect relevance, canonical conflicts, lost internal links, missing content and rendering first.
FAQ
Can a website redesign improve SEO as well as preserve it?
Yes, but sequence matters. First protect URLs, content and internal links that already perform. Then improve weak information architecture, duplicated content, page experience and topic coverage. Combining every change into one unmeasured release makes gains and losses difficult to diagnose.
Should every URL stay the same during a redesign?
No. Keep URLs that remain accurate and useful. Change them when the information architecture or page purpose genuinely requires it. Cosmetic slug clean-ups create risk without much upside, so the case for changing an established URL should be stronger than “the new one looks tidier”.
How long should permanent redirects remain in place?
Google recommends at least one year. Keeping valuable rules longer protects visitors using old bookmarks, backlinks, proposals and email sequences. Review the rules periodically, but do not remove them merely because the redesign project has closed.
How much organic traffic loss is acceptable after launch?
There is no honest universal percentage. A short period of volatility can occur, especially when many URLs or page contents change. Set thresholds from the site's normal weekly variation and seasonality. Investigate immediately when priority pages become non-indexable, redirects fail or qualified organic leads fall outside that range.
Summary
- Preserve proven search demand before trying to improve it
- Build one evidence-based inventory with an owner and action for every URL
- Map replacements by search intent, not merely by subject or navigation label
- Use direct permanent redirects and update every internal signal to the final URL
- Apply pass-or-fail QA, then monitor rankings and qualified leads by page group
Actualyse designs and rebuilds B2B websites that turn research visits into qualified pipeline. Book a call to talk through where yours stands.
