website redesign seo
No URL changes, no platform change, and traffic drops anyway. Redesigns lose traffic through content that got shortened, links that got removed and headings that became images.
Primary riskEverything that carried ranking signals lives in templates, and a redesign rebuilds every template at once.
A redesign is presented as a visual project. Same URLs, same platform, same content — just a new look. It rarely gets a migration plan, and it rarely gets pre-launch SEO QA.
Then traffic declines over the following six weeks and nobody can point at a cause, because by every measure people are checking, nothing changed.
What changed is everything that carries ranking signals. Rankings do not attach to URLs; they attach to what is at those URLs. A redesign rebuilds every template on the site, and templates are where the copy volume, the heading structure, the internal links and the structured data live.
Design decisions are made in design vocabulary, and the SEO consequence is invisible in that vocabulary. The table on this page maps the translation. Three worth expanding.
“Cleaner and less cluttered” almost always means less text. A category page that carried 600 words of descriptive copy now carries 150, because the new layout has room for 150. That is a real reduction in topical coverage, decided in a design review where nobody was thinking about topical coverage.
“Simplified navigation” means fewer internal links. Footers get trimmed from eighty links to twelve. Sidebars disappear. Related-content modules are cut as clutter. Every one of those was a crawl path and a distribution of internal link equity to deep pages.
“Progressive disclosure” means content behind interactions. That is fine when the content is in the HTML and hidden with CSS. It is a problem when it is fetched on click, because then it may not exist for a crawler at all.
This is the single highest-leverage recommendation on this page.
At wireframe stage, “this template needs space for 400 words and a related-links module” is a note. It costs nothing and nobody argues.
At built-design stage, the same note is a rework request against a signed-off design, and it becomes a negotiation between two teams with different objectives. It usually loses.
The requirements are not complicated: room for the copy that currently exists, a real text H1, retained internal linking modules, and structured data carried over. Getting them into the brief is worth more than any post-launch audit.
Redesigns get bundled with replatforms, URL changes and domain changes because it feels efficient to do everything at once.
It is efficient right up until traffic moves, at which point you have four simultaneous changes and no way to attribute the effect to any of them.
Where the timeline allows, separate them and confirm recovery between each. Where it does not — and often it genuinely does not — the mitigation is a much stricter parity process, since you lose the ability to isolate variables after the fact.
If the platform is changing too, replatform is the broader page. If URLs are changing, URL structure change covers the mapping work. The parity method that underpins all of these is set out on the technical scope page.
What you see Category or service pages lose half their copy because the new design has less room. Rankings decline gradually over the following weeks.
What fixes it Diff word counts per URL before launch. Where copy must be shortened, do it as an editorial decision with the traffic exposure known.
What you see Deep pages lose rankings while the homepage is untouched. The new templates simply link to fewer places.
What fixes it Compare inlink counts per URL between the old and new crawls. Rebuild or replace the modules that carried the links.
What you see The visible heading is now an image or a styled div, and the actual H1 is the logo in the header on every page.
What fixes it Verify heading structure in the parity diff. Text headings, marked up as headings, on every template.
What you see Rich results disappear weeks after launch, with nothing in the traffic data pointing at a cause.
What fixes it Inventory structured data by template beforehand and treat rebuilding it as part of the template work rather than a follow-up ticket.
| Design decision | What it actually changes | Consequence | Detection |
|---|---|---|---|
| "Cleaner, less cluttered pages" | Body copy is cut or moved behind interactions | Word count drops; topical coverage thins | Compare word counts per URL between crawls |
| "Simplified navigation" | Internal links removed from headers, footers and sidebars | Deep pages lose link equity and get crawled less | Compare internal link counts per URL |
| "Modern hero sections" | The H1 becomes an image, or moves below decorative markup | The primary topical signal weakens or disappears | Parity diff on H1 presence and content |
| "Progressive disclosure" | Content moves into tabs or accordions that load on click | Content may not exist in the initial HTML at all | Fetch raw HTML and compare against rendered DOM |
| "Consistent imagery" | Images replaced; alt text not carried over | Image search traffic disappears; accessibility regresses | Compare alt attribute coverage between crawls |
| "Streamlined footer" | Sitewide links to deep sections removed | Crawl paths to those sections narrow considerably | Compare crawl depth per URL |
Same method as a replatform, smaller scope. Everything here compares the old templates against the new ones.
Full crawl with per-URL word counts, heading structure, internal link counts, structured data types and image alt coverage recorded.
Output Pre-redesign crawl baseline
At wireframe stage, not at build. A design that has no room for body copy is much cheaper to change before it is built.
Output Design review notes
Grouped by template rather than by URL, so a systematic reduction is visible as a pattern rather than as scattered individual differences.
Output Word count comparison by template
Both outbound links per page and inbound links per URL. The second is what reveals which pages lost support from navigation changes.
Output Internal link comparison
One H1 per page, real text, logical nesting. Structured data present on every template that had it before.
Output Heading and markup verification
Tabs, accordions and lazy-loaded sections — confirm their content exists in the initial HTML response rather than appearing only on interaction.
Output Content availability report
Yes, and it is one of the more common ways sites lose traffic without an obvious cause. Rankings depend on content, headings, internal links and markup — all of which live in templates. A redesign rebuilds every template, so all of it is in scope even though no address changed.
Reducing copy in the name of a cleaner layout. Design reviews assess how pages look, not how much they say. A category page that drops from 600 words to 150 loses topical coverage, and the loss appears weeks later with no obvious trigger.
At wireframe stage. Once designs are built, changing them is expensive and the conversation becomes adversarial. At wireframe stage, "this template needs room for 400 words and a related-links module" is a cheap, uncontroversial note.
Only if the content is not in the initial HTML. Content hidden behind a tab but present in the HTML response is generally fine. Content fetched when the tab is clicked may never be seen by a crawler at all. Check the raw response rather than assuming.
Send the launch date and what is changing. We will tell you which parts of this page apply to your situation and which do not.