Website redesign for SEO
A poorly executed redesign drops organic traffic by 30 to 40%. The site looks better, nothing errors, and the traffic leaves quietly over about eight weeks. It is almost never the design that causes it. It is nine specific, boring, preventable things, and every one of them is cheaper to check before launch than to diagnose after.
Why it happens quietly
A redesign does not break search visibility the way a server outage breaks a site. Nothing returns an error. Nobody gets an alert. The pages still load, and they look better than they did.
What changes is whether search engines can still find, read and match your pages to the queries they used to match. That degrades over weeks, not minutes, because a crawler has to come back, re-read and re-evaluate before anything moves.
By the time a report shows it, the developer has moved on and the cause is four weeks behind you.
The nine things that actually cause it
Ranked by how often I find them, not by how dramatic they sound.
1. A noindex left over from staging. The single most expensive line of code in web development. The staging site was hidden from search, the flag shipped with the launch, and the entire site disappears within days. Check the live site, not the code.
2. URLs changed without redirects. Every address that moves needs a permanent redirect to its closest equivalent. Not to the homepage. A redirect to the homepage tells a search engine the old page is gone, and it takes the ranking with it.
3. Internal links lost in the navigation rebuild. This is the one I have lived. A page nothing links to is invisible regardless of how good it is. Rebuilt navigations often swap real links for anchors or JavaScript handlers, and every page that was only reachable from that menu goes dark.
4. Content removed during simplification. The 900-word page that ranked gets condensed to 200 words of headline and whitespace. It looks cleaner and it now answers nothing, so it matches nothing.
5. Titles and headings rewritten for aesthetics."Website maintenance services" becomes "Keeping things running". Better rhythm, and it no longer contains the words anyone searches for.
6. Redirect chains. Old URL points to an interim URL that points to the new one. Each hop loses a little and slows the crawl. Every redirect should reach its destination in one step.
7. The sitemap not updated. It still lists URLs that now 404, and omits the new ones. Search engines treat a sitemap that lies as a sitemap not worth reading.
8. The new site is slower. Heavier images, more scripts, a framework doing more than the old page needed. Speed is a smaller ranking factor than people claim, but it compounds with everything else on this list.
9. Structured data dropped. Whatever produced your rich results in search quietly stops existing, and the listing loses whatever made it stand out.
What to check before you launch
Nine items, in the order I run them. None of them takes long.
| Check | How |
|---|---|
| No noindex anywhere | Request the live page and read the source. Not the repository |
| Every old URL resolves | List every indexed URL, request each on the new site, expect 200 or a single 301 |
| No redirect chains | Follow each redirect to the end and count the hops. One |
| Every page has an inbound link | Crawl from the homepage and compare against the sitemap. Anything in one and not the other is orphaned |
| Sitemap matches reality | Every URL in it returns 200, every live URL is in it |
| robots.txt blocks nothing important | Read it. It is four lines |
| Titles and H1s kept their words | Diff old against new. Rewriting is fine; losing the search term is not |
| Structured data survived | Whatever produced rich results before should still validate |
| Speed on a real device | Not a lab score in a report |
The list is boring on purpose. Every item on it is something I have found broken on a site that was already live.
What to measure after
Week 1. Coverage errors in Search Console. New 404s are the fastest signal that a redirect was missed.
Weeks 2 to 4. Impressions, not clicks. Impressions move first because they reflect what search engines are willing to show. A drop here is the early warning; clicks confirm it a fortnight later.
Weeks 4 to 8. Pages with zero impressions. Any page that had traffic before and has none now is orphaned, blocked, or no longer matching. All three are fixable, and all three are invisible unless you look.
If nothing has moved by week eight, the migration was clean.
The same mechanism, running the other way
I rebuilt matted.co.uk, a service business in London. Search Console impressions went from 0 to 46,800 in three months.
The site was never penalized. It was structured in a way that let nothing find it, and fixing the structure is what produced the number. Same mechanism as everything above, pointed in the opposite direction.
And I have been on the wrong side of it. Two pages on my own site sat invisible for two months because the navigation used anchors instead of real links. Not a penalty, not thin content. Nothing could reach them, so nothing did.
That is why the checklist above is nine boring items rather than a paragraph about best practices. Every one of them is a way I or someone I have audited has lost traffic.
If you are planning one
The cost side is in website redesign cost. If your site sells to other businesses, the buying committee changes what gets built: B2B website redesign.
And if you would rather someone ran the nine checks against your current site before you commit to anything, that is what the website audit is for.
More questions about website redesign are answered on the topic page: when one is worth doing at all, how long it takes, and what happens to a site while it is being rebuilt.
The 30 to 40% migration figure is drawn from Orbix Studio and Digital Applied. The 0 to 46,800 on matted.co.uk is mine, measured on 3 August 2026, and so are the two months my own pages spent unreachable.