Migration problems rarely announce themselves as one event. Traffic may hold for a week, then slide as crawlers reprocess the site. A launch that looks clean in a browser can still ship a noindex tag, a redirect loop or an analytics script that never fires. This guide is for the team past launch, or about to cross it, that needs to diagnose what changed and fix it in order.
The first rule is to stop changing things blindly. Compare the live site with the pre-migration baseline, isolate one cause, fix it and record the date. Repair the chain from the crawl layer upward.
Start with a 48-hour diagnostic
Collect the same evidence you should have captured before launch: indexed URL count, organic landing pages, crawl errors, redirect coverage, sitemap status, page speed and conversion events. Use Search Console, a crawler pointed at the new domain, server logs and analytics. Fix in sequence: crawling first, then ranking signals such as titles, canonicals and internal links, then conversion issues such as checkout and analytics. Changing all three at once makes it impossible to know which fix worked.
Symptom: organic traffic falls within days of launch
Likely causes. A staging noindex tag copied to production, a robots.txt block, a firewall or basic authentication prompt, a DNS change pointing at the wrong host, or canonical tags still referencing the old domain.
Fix. Fetch the production pages from outside the office network and inspect the response headers, robots.txt, meta robots and canonical. Confirm the DNS record and the certificate. Ask a colleague on a mobile connection to open the site and a deep page, not just the home page.
Prevention. Add a launch gate that checks these four signals on the live host, not on staging. A launch that only passes in the development environment has not been tested.
Symptom: pages stay indexed but rankings and clicks fall
Likely causes. The new template regenerated titles, descriptions and headings, structured data was dropped, or the URL structure changed the meaning of the page. A redesign can also remove internal links that told search engines which pages mattered.
Fix. Restore the closest equivalent metadata for pages with traffic and backlinks, revalidate structured data and check that the primary heading still describes the page. Rebuild the main navigation and contextual links before adding new content.
Prevention. Keep metadata in the migration map rather than in a separate content project. The person who owns the redirect map should also own the title and description map.
Symptom: new URLs are slow to index
Likely causes. The sitemap still lists old URLs, internal links point to redirects, the canonical points elsewhere, or the site returns inconsistent status codes for the same page.
Fix. Publish a current sitemap, update navigation and footer links in the templates, make every old URL resolve in one hop, and use the change of address tool where the domain changed. Submit canonical URLs, not redirecting ones.
Prevention. Crawl the new site as an anonymous user before launch and again a week after. Check the crawl report for redirects and 5xx responses, not only for 404s.
Symptom: redirect chains, loops and soft 404s
Likely causes. The CMS adds a trailing slash, the load balancer adds or removes www, HTTP and HTTPS both redirect, or a catch-all rule sends every missing page to the home page.
Fix. Build a single-hop map for each old URL, test both protocol and host variants, and return a real 404 or 410 for retired pages. A redirect to the home page is not a relevant destination.
Prevention. Keep the redirect map in version control and review it whenever a template, language or store view changes. Redirect rules decay quietly when nobody owns them.
Symptom: traffic lands on the wrong page
Likely causes. Broad redirect rules, language or region targeting mistakes, or pagination and filter URLs collapsed into one page. A separate Bangla and English structure adds another way to send a visitor to the wrong language.
Fix. Review the top landing pages by sessions and conversions, compare the old and new intent, and correct the closest match. Preserve query strings that change content, such as filters and pagination. Check hreflang and language selectors after the fix.
Prevention. Assign a reviewer who did not build the map to open a sample of fifty redirects and judge whether the destination is genuinely equivalent.
Symptom: Core Web Vitals regress
Likely causes. A heavier template, new third-party scripts, unoptimised images, a different font strategy or missing caching. A migration often adds a page builder, a chat widget and three tracking scripts in the same release.
Fix. Measure the main templates on a mobile connection, remove or defer non-critical scripts, serve modern image formats and restore the previous caching rules. Test on a mid-range Android phone, not only on a desktop.
Prevention. Put performance budgets in the launch gate. A template that adds a second of load time to every product page is a business change, not a design preference.
Symptom: conversions fall while traffic holds
Likely causes. Checkout or forms broken by the new theme, cookie consent blocking analytics, a payment provider callback mismatch, or locale and currency defaults changed.
Fix. Run a real order or enquiry end to end on mobile, confirm that events reach analytics, and test the payment sandbox and production callbacks separately. Check that confirmation emails and SMS still send.
Prevention. A conversion smoke test belongs in every migration checklist. Marketing traffic that reaches a broken form is more expensive than no traffic at all.
Symptom: backlinks point to dead URLs
Likely causes. The map covered pages but not files, feeds, campaign URLs or partner links. External sites linked to the old paths and nobody checked the referring URLs.
Fix. Identify the top linked URLs, add single-hop redirects and contact the few partners whose links broke. Prioritise links that send real referral traffic over links that only affect a vanity metric.
Prevention. Include backlink data in the baseline and recheck it after thirty days. A migration is not finished when the new site launches; it is finished when the old paths have stopped losing value.
Bangladesh-specific checks
Test on mobile networks, confirm that Bangla text renders correctly in titles and snippets, check that local payment providers still complete, and verify the .com.bd domain and DNS records if the domain changed. A migration can be technically clean and still fail a customer on a slow connection.
Keep a daily dashboard for indexing, organic sessions, conversion events, server errors, redirect status and uptime for at least thirty days. Review it weekly with marketing, development and support, and close the watch only when the trend has returned to the baseline or the remaining gap has an explained cause.
For the pre-launch version, read the pre-launch migration checklist. If you need the hosting, DNS and monitoring side handled together, start with Managed Technology & Cybersecurity; the pricing page explains how the work is scoped.
Sources
- Google Search Central, Site moves with URL changes, accessed 17 November 2026.
- Google Search Console Help, Change of address tool, accessed 17 November 2026.
- MDN Web Docs, 301 Moved Permanently, accessed 17 November 2026.
- web.dev, Web Vitals, accessed 17 November 2026.
Responses (0)
No comments yet. Add the first one below.
Leave a response
Your email address will not be published. Required fields are marked *.