A redesign that loses traffic almost never loses it because the design changed. It loses it because pages moved, pages got shorter, or something technical came across from the staging site.
All three are preventable, and all three are far cheaper to prevent than to diagnose two months later.
The three things that cause it
URLs changed and nothing sends the old ones anywhere. Every link anyone has ever made to you, and every result a search engine is currently holding, now lands on a missing page. This is the big one and it's the easiest to avoid.
The content got shorter. A cleaner design usually means less text, and the page that ranked was often the long ugly one. Ranking follows what a page answers, so a page that now answers less of the question ranks for less of it.
Something technical came across from staging. A noindex tag or a robots.txt rule. Correct while the site was being built, costly the moment it went live. Why a site isn't showing up on Google is the diagnosis for that one.
Start with an inventory, before anything is designed
Before a single page is laid out, you want two lists.
- Every URL the current site has, from the sitemap, from the CMS, and from a crawl, because none of them are complete on their own.
- Every URL that earns anything. Search Console, last twelve months, sorted by clicks and then again by impressions. That second sort matters: a page with impressions and no clicks is ranking, and it's worth keeping.
Most sites turn up pages nobody remembers making that bring in work steadily. Deleting one of those during a tidy-up is the most common self-inflicted wound in a redesign.
Content parity on the pages that earn
For any page on that list, the new version has to answer at least as much as the old one did. It can say it better, in fewer words, with a far better layout, but it can't say less.
This is where a redesign and good taste collide. The instinct to cut a wall of text is usually right for a visitor and wrong for a page that ranks because of that text. Rewrite it, structure it, put half of it behind headings people can skip. Do not delete it.
The redirect map
One row per existing URL: where it's now, where it will be, and whether it's unchanged or redirected. Built before the new site, not after it.
- Permanent redirects, so the value transfers. A temporary one tells search engines to keep the old address.
- No chains. A pointing at B pointing at C loses a little at every hop and eventually gets ignored. Point everything at the final address.
- Never send everything to the homepage. It gets treated as a missing page anyway, and it also annoys the person who clicked.
This is the substance of a redesign done properly and of moving off WordPress, and it's the step that gets skipped when a launch date slips.
Launch day
- The noindex tag is off, checked in the live page source instead of in a settings screen.
- robots.txt allows crawling.
- The sitemap is regenerated and resubmitted in Search Console.
- Redirects tested by loading the old addresses, at least the top fifty by traffic.
- Analytics is running and recording. A redesign that loses its tracking can't prove anything afterwards in either direction.
- SSL works on the www and the non-www version, with one redirecting to the other.
- Every form submitted end to end, from a phone, with the email actually arriving.
The first month
A small dip in the two weeks after a launch is normal while search engines recrawl and re-evaluate. It usually recovers without anyone doing anything.
What's not normal is a cliff, or a dip still deepening after three weeks. Check Search Console weekly: coverage errors first, then your top pages against the before list. When a page that used to bring in work has vanished, it's nearly always a redirect that didn't fire or content that got trimmed.
Export the old site's analytics and Search Console data before you switch. Without a before, a seasonal dip and a self-inflicted one look identical.