Most traffic lost in a website migration is lost to a handful of avoidable mistakes, and almost all of them are decided before launch day. The work that protects rankings is unglamorous mapping done in advance, rather than anything clever done afterwards.
The single most common cause is a redirect map that was never built, because the assumption was that URLs would stay the same. They rarely do.
Capture the before picture
- Export every indexed URL, not just the ones in the navigation.
- Record current rankings and organic traffic for your top pages, so a drop can be measured rather than argued about.
- Export the internal link structure, because that is what you will rebuild.
- Note which pages have earned external links, since those are the ones that must not break.
That fourth point decides priority. A page with genuine inbound links is worth more than its traffic suggests, and losing it costs more than losing a page nobody references.
Map every old URL to a new one
One to one wherever possible. Redirecting everything to the homepage is the classic destructive shortcut: it preserves nothing and tells search engines the old page no longer has an equivalent. Where a page genuinely has no successor, redirect to the closest relevant section rather than to the root.
Use permanent redirects, and avoid chains where one URL redirects to another which redirects again. Chains dilute the signal and slow the page, and they accumulate silently across successive rebuilds.
Nine out of ten migrations that lost traffic did not have a redirect map. The tenth had one and nobody tested it after launch.
Daniel Okafor, Director of SEO, Media @ Marsons
Freeze the things that do not need to change
A migration is a tempting moment to change everything at once. Resist it. Changing the design, the URL structure, the content and the platform simultaneously means that when traffic moves, nobody can say which change caused it. Where the structure is sound, keep the URLs and change the rest.
Check the staging site is not indexable
A staging site left open to crawlers creates duplicate versions of every page, and the cleanup is worse than the prevention. Block it, and then confirm the block was removed before launch, because a live site still carrying its staging restrictions is the other half of this mistake and it is remarkably common.
Launch day checks
- Spot-check redirects from the real old URLs, including deep pages nobody remembers.
- Confirm the live site is crawlable and the staging block is gone.
- Check analytics and conversion tracking still fire, since tags are frequently lost in a rebuild.
- Submit the new sitemap and confirm it lists the URLs you expect.
- Verify the canonical tags point where you intend rather than at the old domain.
The dip that is normal
Expect some volatility for a few weeks while the new URLs are recrawled and reassessed. A modest dip that recovers is ordinary. A sharp drop that does not recover within a month is a signal to check the redirect map and the indexability rather than to wait longer, because the usual cause is mechanical and fixable.
Keep the old redirects forever
Redirect rules from previous migrations are frequently deleted during the next one, which quietly breaks links that had been working for years. Keep them, consolidate chains where they have formed, and treat the redirect file as a permanent asset rather than as launch scaffolding.
For how site structure affects what can rank in the first place, the SEO hub covers the market and sector pages.
Who needs to be in the room
Migrations fail at the handovers between people rather than inside any one discipline. The developer knows the URL structure is changing and does not know which pages carry external links. The marketing lead knows which pages matter and does not see the deployment plan. Whoever owns analytics finds out afterwards that the tags were rebuilt without the conversion events.
One person should hold the checklist and confirm each item with the person responsible for it. That role is unglamorous and it is the difference between a migration that goes quietly and one that produces a month of forensic work. Where an agency and an internal team are both involved, agree explicitly which of them owns the redirect map, because assuming the other side has it is the single most common failure in this whole process.
Do not migrate into a busy season
Launching a rebuild the week before your busiest trading period concentrates all the risk at the point where it costs most. Where the business has a seasonal peak, migrate well before or well after it, and accept that a delay of a few weeks is far cheaper than a recovery in the middle of it.
The same applies to launching on a Friday. Give yourself working days after go-live with the people who built it still available, since the issues that matter usually surface in the first forty-eight hours.
Planning a rebuild or a domain change?
We will map the redirects and run the pre and post-launch checks, so the new site keeps the visibility the old one earned.
