Website Relaunch

I start with a baseline audit. I pull the last three months of Search Console clicks, impressions, and average position. I export analytics data for organic sessions, bounce rate, and conversion rate. Then I run a full crawl of the current site using Screaming Frog, documenting every URL, status code, meta data, and internal link. I also check backlink profiles via Ahrefs to identify pages with valuable external links.

Next, I map every URL to its new destination. For pages with strong signals — rankings, traffic, backlinks — I keep the URL if possible. If the structure must change, I create a direct [301 redirect](/301-redirect/) to the closest relevant page. I avoid redirect chains and ensure the target page contains equivalent content. I build the redirect map in a spreadsheet with three columns: old URL, new URL, and status (keep, redirect, remove). For removed pages, I redirect to a parent category or a custom 410 if truly gone.

I test the entire redirect map on a staging environment by modifying the hosts file to point to the staging server. I check forms, analytics tracking, and rendering of key pages. I pay special attention to pagination, [canonical tags](/canonical-tag/), and hreflang if applicable. Once satisfied, I update the DNS and submit the updated XML [sitemap](/sitemap/) to Google Search Console. I monitor Search Console for crawl errors, 404s, and indexing rates daily for at least two weeks.

Example: On a recent relaunch of a 2,000-page e-commerce site, I identified 150 high-value pages that I kept unchanged. For the remaining 1,850, I created redirects. Post-launch traffic dropped 5% in the first week but recovered to pre-launch levels by week three because I had mapped every URL correctly. I also implemented performance improvements like image optimisation and lazy loading, which improved Core Web Vitals and likely helped retention.

The biggest risk is incomplete redirect mapping. I saw a colleague lose 40% of traffic because they forgot to redirect old blog posts that had backlinks from high-authority sites. My rule: inventory every URL, even those with zero traffic.

Also, do not launch without staging. I once found that a staging environment was blocking Googlebot with [robots.txt](/robots-txt/), and I nearly launched with that setting. Check robots.txt, noindex tags, and X-Robots-Tag headers in staging. Another watch-out: internal links. After launch, old internal links will point to URLs that now redirect, creating unnecessary chains. Update internal links in content, navigation, and sitemaps. Use a database search and replace for CMS content.

For [JavaScript-heavy sites](/javascript-seo/), ensure pre-rendering or server-side rendering so Google can see the content. I have seen relaunches where the new site uses a JavaScript framework that takes 5 seconds to render, causing poor indexing. Test with Google's URL Inspection tool and Mobile-Friendly Test.

Counter-argument: Some argue for a 'big bang' launch to minimise dual maintenance. I prefer phased launches for sites over 1,000 pages. Launch the homepage and top categories first, then subcategories, then product pages over a week. This allows me to catch issues early. However, phased launches require careful canonical management to avoid duplicate content.

Finally, track changes using a scorecard: pre-launch baseline vs. first month. Measure clicks, impressions, rankings, conversion rate. If traffic drops more than 10% after two weeks, investigate specific sections. Example: In one relaunch, I found that a category page's canonical tag pointed to itself but the old URL had backlinks. The redirect was correct, but the canonical was wrong. I fixed it and traffic recovered in four days.

I have made almost every mistake in the book. The worst was a relaunch in 2019 for a 5,000-page news site. I thought I had mapped all URLs, but I missed a section of archived articles. Those pages had thousands of backlinks from external news aggregators. Traffic to that section dropped to zero. Over two months, organic traffic dropped by 35%. I had to re-redirect those old URLs, but some equity was lost. Now I use Screaming Frog to crawl the old site and compare all URLs against the new site's crawl, flagging any that are missing. I also check server logs for 404s post-launch.

Another mistake: not testing forms. After one relaunch, the contact form sent emails to a non-existent address. I lost a week of leads. Now I create a test submission in staging and verify the email arrives.

I wrongly assumed that a 301 redirect passes all PageRank. Google has said that a 301 passes most, but not all. In practice, I have seen a 20% drop in rankings for redirected pages that had strong anchor text. For high-value pages, I now keep the URL if at all possible, even if it means a less ideal information architecture. Over time, Google may adjust, but preserving the URL avoids the risk.

Finally, I underestimated the impact of canonical tags. In one relaunch, I set the new URLs as canonicals but forgot to set self-referencing canonicals on the new pages. Google indexed both old and new as duplicates, splitting signals. Fixed it, and rankings consolidated in two weeks.