How to redesign a website without losing your Google rankings

A new website is meant to bring in more work. Done carelessly, it does the opposite: the old pages that Google had learned to trust disappear, the links other sites pointed at them break, and enquiries dry up for months while nobody can work out why.

I’ve done this migration for clients such as KPM First Aid Training, and most recently for the SR7 website you’re reading now, which moved from WordPress to a hand-built static site. This is the process, with real numbers from that rebuild.

Why rankings drop after a redesign

Google doesn’t rank “your website”. It ranks individual pages, at individual addresses. When a redesign changes those addresses, or removes the words that made a page relevant, Google has to work everything out again. The usual causes are:

• Changed URLs with no redirects. /our-services/ becomes /services/, and every result and link pointing at the old one now leads to a 404.
• Deleted pages that were quietly doing a job. An old blog post or a town page nobody on the team valued turns out to have been bringing in a steady trickle of searches.
• Copy cut for looks. A slick new design with a third of the words can lose the very phrases a page ranked for.
• A staging site that goes live with noindex still switched on. It happens more often than anyone admits.

Step 1: record everything before you change anything

Before designing a single page, I crawl the whole of the existing site and write down every address, title, description, heading, image and link. For the SR7 rebuild that crawl found 93 live addresses, 33 blog posts, 11 case studies and 326 “attachment” URLs that WordPress had generated for images, which most people don’t know exist.

Then I pull the numbers from Google Search Console: which pages earn impressions and clicks, which searches they appear for, and which pages other websites link to. For SR7 that showed:

• 309 clicks and 14,900 impressions over 16 months;
• every one of the site’s 51 external links pointing at the homepage;
• a deleted post, still returning a 404, that had earned 157 impressions and 3 clicks;
• large numbers of impressions for Durham and Sunderland searches, at positions 30 to 50.

That record is what the new site is checked against at the end. Without it, you’re guessing.

Step 2: keep the addresses that already work

The single most effective thing you can do is keep existing URLs exactly as they are wherever possible. It’s tempting to “tidy up” slugs so they match a new structure. Resist it. An address with years of history is worth more than a neater-looking one.

On the SR7 rebuild every page, post, case study and category kept its original address. New pages were added alongside them rather than replacing them.

Step 3: write the redirect map

Where an address genuinely has to change, or a page is being removed, it needs a permanent (301) redirect to the closest matching page. A good redirect map has three rules:

1. Relevant destinations. A removed page about roof repairs redirects to the roof repairs page, not the homepage. Sending everything to the homepage is treated by Google as a “soft 404” and passes on very little.
2. No chains. Old page → new page, directly. Not old page → interim page → new page.
3. No loops. Obvious, but easy to create by accident with broad rules.

For SR7, the 326 image attachment URLs had been redirecting to the homepage, which is exactly the soft-404 pattern to avoid. In the rebuild each one now points to the page the image actually belongs to. The deleted post redirects to the journal, and an old GDPR statement redirects to the privacy policy.

Step 4: carry the valuable content across

Copy can be improved, but anything that earns search traffic should keep its subject and its substance. On SR7 every case study kept its full text, every post kept its title and description, and the pages with the most impressions gained content rather than losing it: the Durham and Sunderland pages picked up new, genuinely local sections because Search Console showed they were close to being useful.

Step 5: the technical checks before launch

• Every page has one canonical address, and it matches the live URL.
• No page is accidentally marked noindex, and robots.txt isn’t blocking the site.
• The XML sitemap lists every page you want found, and nothing you don’t.
• Structured data, titles, descriptions and headings are present on every page.
• Internal links point straight at final addresses, not at redirects.
• The Search Console verification code is carried over, so you don’t lose access to your own data.
• Analytics is working, and enquiries are tracked as conversions.

I automate most of this: a script checks every page of the new build against the old crawl, fails if any old address is neither live nor redirected, and flags broken links, missing images and duplicate titles.

Step 6: after launch

Switch over in one step, then submit the new sitemap in Search Console and request indexing for the most important pages. For the next few weeks, watch the Pages report for new 404s or redirect errors and fix them quickly. Some movement in positions is normal while Google recrawls; what matters is that the trend recovers and then improves.

When you shouldn’t redesign at all

Sometimes the honest answer is that the site doesn’t need rebuilding. If it’s on a sensible platform and the problems are a slow image or two and a dated homepage, targeted fixes are cheaper and carry no migration risk. A free website review is a good way to find out which you need.

In short

Record everything first. Keep the addresses that work. Redirect what must change to the closest relevant page, directly. Keep the content that earns traffic. Check the technical basics before launch, then watch closely afterwards. If you’d like a redesign done that way, here’s how I approach it.