Redesigning a Website Without Losing Traffic

A redesign fails in two visible ways: it looks wrong, or traffic falls off a cliff. The second one is worse, because it usually arrives three weeks after launch, when the team that built it has moved on and nobody can tell what caused it.

The cause is almost never the design. It is that the old site's structure was never recorded, so the new site could not be told what to preserve. Everything below is cheap before launch and expensive after.

Before anything is deleted

Export the list of every URL that currently exists. Not the pages you remember — every URL. Crawl the site or pull the sitemap, then cross-check against Search Console's indexed pages, because the two never agree and the difference is where trouble hides.

Mark which of those URLs earn anything. Two sources: organic landings in analytics, and the queries people arrive on in Search Console. On most small business sites, a handful of pages carry nearly all of the value, and they are rarely the pages the owner is most attached to.

Record the current content of the important pages. Copy, headings, structured data, the exact wording of a service description that happens to rank. When a page comes back rewritten, you need to be able to tell the difference between an improvement and an accident.

Note every redirect that already exists. Old campaign URLs, shortened links, print references, things fixed two redesigns ago. These chain silently, and a redirect pointing at a page that is about to be deleted becomes a redirect pointing at a 404.

Screenshot the site. Boring, and the only thing that settles an argument six weeks later about what was there before.

The decisions that actually move rankings

Keep the URLs you can keep. Every changed URL is a page that has to be re-learned. If a page survives the redesign, it should keep its address unless the address itself is the problem. A tidy URL structure is not worth losing a page that ranks for a term with commercial intent.

Map every change by hand. A spreadsheet: old URL, new URL, reason. Then implement redirects from that sheet, not from memory. Anything left unmapped becomes a soft 404 that Search Console will report weeks later as a drop.

Preserve intent, not just pages. If a page ranks for "home tuition in [area]" and the new structure folds that into a general services page, the redirect is technically correct and the ranking is gone — because the destination no longer answers the query. When a page's purpose disappears, the page should survive as its own URL.

Do not merge pages because there are too many. Consolidation is a legitimate strategy and the most common way a redesign quietly loses a month of traffic. Merge when two pages genuinely answer the same question. Merge for tidiness and you delete the one that was ranking.

What breaks silently

Structured data. Type, service and local-business markup gets dropped in the rebuild because it lives in templates nobody reviewed. It rarely produces an error — your listing just gets less useful and clicks fall.

Heading structure. A page whose H1 was "Physiotherapy for Sports Injuries" comes back with that text inside a decorative banner and the H1 set to "Welcome". Search reads the second version.

Images. New filenames, no alt text, and a hero image that adds a second to load on a phone. Alt text describes what is in the image; it is not a place to put a keyword nobody would search.

The enquiry path. Forms, WhatsApp links and phone numbers that worked on the old site and were never re-tested on the new one. This is the failure that produces no warning at all — traffic looks normal while enquiries stop.

Tracking. Analytics and conversion events pointing at page names that no longer exist, so the report says nothing changed.

Canonical tags. Left over from a staging environment, or pointing at old URLs. This one can make an entire new site unindexable while looking perfect in a browser.

Launch day

Submit the new sitemap. Verify Search Console access survived the move. Test every redirect from the mapping sheet, not a sample. Check that the important pages return 200 and one canonical each. Confirm robots.txt is not accidentally blocking the whole site — a staging rule left in place is a classic way to lose everything for a fortnight.

Then leave it alone. The temptation on day two is to fix a layout nitpick, which touches templates, which is exactly when something breaks during the observation window.

The six weeks after

Expect movement. Rankings shift during re-crawl and settle; a page that drops in week one and returns in week four has not necessarily been damaged.

Watch three things weekly: indexed page count, impressions on the pages that used to earn, and enquiries. Click-through rate tells you the titles and descriptions are working. Enquiries tell you the page still does its job. If traffic holds and enquiries fall, the problem is the page's structure or its call to action, not search.

Fix in order: unmapped URLs returning 404, pages that lost their structured data, pages whose content drifted away from the query they ranked for, then everything cosmetic.

When a redesign is the wrong answer

If the current site ranks and converts, a redesign is a risk you take for reasons that are usually about pride. Fix the specific page that is failing, improve the slow templates, sharpen the enquiry path. Redesign when the structure is genuinely wrong — when the site is organised around how the business thinks rather than how customers look, or when adding a page requires asking a developer.

That conversation is worth having before anyone quotes, because "redesign" and "fix these four pages" are very different invoices.