Redesigning a website that already ranks is a different job from building a new one. The pages have history, the URLs have links pointing at them, and every slug you rename is a decision Google has to be told about. Get that part wrong and the new site looks better while performing worse.
This is a worked example of how we handle it, using the rebuild of Ideal Jewelry & Loan, a family-run pawn shop and gold buyer in Brockton, Massachusetts. It covers the URL audit, the structure we shipped, and one redirect trap that makes a broken migration look healthy to anyone spot-checking it. If you are planning a website redesign or rescue, the middle section is the part worth reading twice.
Key Takeaways
- A 301 redirect is not proof a migration worked. WordPress guesses a destination for unknown URLs, so a renamed page can appear to redirect correctly while landing on the wrong page entirely.
- Deleted categories and tags usually return 200 with an empty page, not 404. A status-code check marks them healthy; Google keeps them indexed as thin pages.
- Exact-match redirect rules can miss real traffic, because live visitors arrive with
utm_andgclidparameters attached. - Ideal Jewelry & Loan went from 23 pages to 33, with service, location and staff pages separated out rather than combined.
Who Is Ideal Jewelry & Loan?
Ideal Jewelry & Loan is a pawn shop, gold buyer and jewelry store in Brockton, Massachusetts. Their own site describes the business as family owned since 1955, and they serve the surrounding Massachusetts towns as well as Brockton itself.
That matters for the rebuild, because it is a local business with a long trading history. An established local site carries years of accumulated search presence, directory citations and customer bookmarks. The rebuild had to protect all of it.
What Changed in the Rebuild?
The old site ran on a customised build of the Salient theme. The new one runs on Elementor with the lightweight hello-elementor parent theme, which is the stack we use for most WordPress website design work because it keeps the client editing their own pages without a developer.
The bigger change was structural. The site went from 23 pages to 33, and the additions were deliberate rather than padding:
- One page per service. Pawn loans, gold loans, jewelry loans, instant cash, cash for gold, cash for silver, coin dealing, fine watches, diamonds, luxury goods, bullion, gift cards, jewelry repair and selling items each got a dedicated page instead of sharing one.
- One page per town served. Brockton, Bridgewater, Randolph, Stoughton and Easton, plus an index page covering the service area.
- One page per team member. The old single “Meet the Team” page became individual pages, which suits a business where customers are handing over valuables and want to know who they are dealing with.
- A gold price per gram calculator, giving visitors a reason to arrive and a reason to return.
Splitting a combined page into several only helps when each new page answers a genuinely different question. A separate page for “cash for silver” and “cash for gold” works because those are different searches by different people. Splitting for the sake of page count produces thin content and competing URLs.
How Do You Redesign a Website Without Losing Its Search Rankings?
The short answer: enumerate every URL on the old site, decide what happens to each one, and then verify the result with a tool that reports the truth. The order matters. Most redirect problems come from mapping URLs after launch instead of before.
Map every old URL before anything goes live
We pulled the full page, post, category and tag list from both sites through the WordPress REST API rather than crawling a sitemap. A sitemap only lists what the SEO plugin chose to include, so noindexed pages and taxonomy archives never appear in it, and those are exactly the URLs that quietly break.
On this rebuild that produced a clear map: 7 pages kept their URL, 10 were renamed, and 3 had no direct replacement. Every renamed page needed a redirect, and the 3 without a replacement needed a decision rather than a default.
The trap: a 301 that looks correct and is not
This is the finding worth carrying into any migration.
When WordPress receives a URL it does not recognise, it does not necessarily return 404. A built-in function called redirect_guess_404_permalink() looks for a post whose slug starts with the requested one and sends the visitor there with a 301. That behaviour is on by default.
The consequence is that a completely unconfigured migration can look finished. Check an old URL, see a 301, and conclude the redirect is in place. It is not. WordPress is guessing, and on this site the guess was wrong for 5 of the 10 renamed service pages, each landing on a blog post rather than the service page it should have reached.
That failure is worse than a 404. A 404 tells Google the page is gone. A 301 to the wrong page tells Google the service page’s authority now belongs to an article, and moves it there.
We proved the guesser was responsible rather than assuming it, by requesting /cash-for-sil/, a slug that has never existed on either site. It returned a 301 to a blog post. A genuinely random slug returned a clean 404. That is the signature of the guesser, and it is the check worth running on any WordPress site mid-migration: if a URL you invented redirects instead of 404ing, your redirect audit is measuring nothing.
After we configured the real redirects, those URLs resolve where they should. Requesting /diamond/ now reaches the Diamonds page, /fine-watches/ reaches Fine Watches Services, and /gold-buyer/ reaches the Cash for Gold page.
Soft 404s: the pages a status check calls healthy
Categories and tags that exist on the old site and not the new one rarely return 404. WordPress renders the archive template with nothing in it and returns 200. On this rebuild, 6 categories and 24 tags behaved that way.
A status-code check passes all 30. To find them you have to assert on content, not on the response code: we counted rendered article elements per archive, where a live category returned 10 and each dead one returned 0.
Redirect rules and query strings
One more detail that only shows up under real traffic. An exact-match redirect rule compares the whole requested path, and real visitors do not arrive with clean URLs. They arrive from ads, email and social with utm_source, gclid or fbclid attached.
Without the rule set to ignore parameters, /diamond/?utm_source=google does not match a rule written for /diamond/, and the visitor falls through to a 404. We found this because a verification pass appended a cache-busting parameter and 8 of 13 rules appeared to fail. Re-running without it showed all 13 working, which is what exposed the real problem. A verification that had used clean URLs throughout would have passed and shipped the bug.
What We Would Do Differently
Two things, stated plainly.
The redirect map should be built during the rebuild, not audited at the end. On this project the URL audit came late enough that the gap existed on staging before it was caught. It was caught before launch, which is the point of auditing, but the cheaper place to catch it is at the moment a page is renamed.
Second, staging sites need to be blocked from indexing on the day they are created rather than the day before launch. A staging copy that search engines can reach is a full duplicate of the client’s site competing with the original.
Frequently Asked Questions
Will a website redesign hurt my search rankings?
It can, and the usual cause is URL handling rather than design. If every old URL either keeps its address or redirects to the closest genuine equivalent, a redesign does not have to cost rankings. If URLs change without a redirect map, losses are likely.
How do I check whether my redirects actually work?
Request a URL that has never existed on your site. If it returns 404, your redirect checks are meaningful. If it returns a 301 to some real page, WordPress is guessing destinations and you cannot tell a configured redirect from a guess by looking at status codes alone.
What happens to old blog posts and categories in a rebuild?
Posts usually carry over unchanged. Categories and tags are where content is quietly lost, because removing a term leaves a URL that returns an empty page rather than an error. Decide deliberately whether each one should redirect to a replacement or return a clean 404.
Should a local business have a separate page for every town it serves?
Only where each page can say something specific and true about that location. Five near-identical pages with the town name swapped are thin content. Ideal Jewelry & Loan has separate pages because it genuinely serves distinct Massachusetts towns from one Brockton location.
How long does a rebuild like this take?
It depends on the page count, how much content is being rewritten rather than moved, and how much of the old site needs auditing. A structural rebuild with a full URL map is a different scope from a visual refresh on existing pages, so we scope the two separately.
Planning a Rebuild?
If you have a site that ranks and you are nervous about replacing it, that caution is correct. The design work is the visible part; the URL map is the part that decides whether the new site keeps what the old one earned.
You can see more of our builds in our portfolio, read how we run projects on our process page, or get in touch to talk through what a redesign would involve for your site.
