What kills a migration isn't the redirect map
In a migration, everyone obsesses over the redirect map. And rightly so, it's necessary. But it's almost never the map that kills traffic. What kills it is what the map can't see: decisions made outside the plan by people who don't do SEO, the signals that made a page rank and that a 301 doesn't carry, and the Analytics nobody switched on in time to see the bleeding. The map is the easy part. It's also the one part everyone gets right, and migrations fail anyway.
Everyone who runs a migration obsesses over the redirect map. And rightly so, it's necessary. But across years of running and advising migrations of domain, CMS and URL structure, it was almost never the map that killed the traffic. It was what the map couldn't see.
The redirect map is the easy part
Let's be fair to the map. It's necessary, and it's the heart of any serious migration: a 1-to-1 inventory from old URL to new, permanent 301 redirects with no chains or loops, validation before publishing, monitoring in Search Console after launch. Without it there's no migration, there's an accident. None of this is new. And that's the point.
The map is the part everyone gets right. Step 1 of every checklist, the first slide of every proposal, the thing any agency can do with its eyes closed. If the map were what killed migrations, they'd almost never fail, because the map always gets made. They fail anyway. So the problem lives somewhere else: in everything the map doesn't cover.
What kills it is the decision nobody controls
The worst damage I've seen in a migration didn't come from a badly built map. It came from decisions made outside the plan, by people who don't do SEO, with no warning to anyone. The map maps URLs. It doesn't map decisions.
I watched a client change a page's URL on an external partner's advice and skip the redirect, because "it was just one page". It was the page that held the main keyword, and its traffic dropped by around 80%. I watched another team delete a page halfway through a migration, telling no one, because in the new CMS "it didn't fit". And I watched someone rewrite the content of a page that ranked, gutting the relevance that kept it there, convinced they were just "improving the copy".
No redirect map on earth catches this. The map says where the URL goes. It doesn't say who's allowed to touch the URL, the page or the content, nor what happens when they do. A migration isn't a technical project with a governance problem bolted on. It's a governance project with a technical core. Whoever doesn't decide who decides, loses.
The map maps URLs, not the signals that made the page rank
Here's the most expensive mistake of those who did the redirects all properly and lost traffic anyway. A page didn't rank because it had that URL. It ranked for its content, for the internal links pointing to it, for the authority it built up. The 301 carries part of the URL's equity. It doesn't carry a promise that the destination keeps the content, the links and the intent.
Redirect a strong page to a consolidated, thinner one: you kept the URL's promise and broke the content's. Rebuild the architecture without recreating the internal-link graph that fed those pages: you scattered the signals across a site that no longer concentrates them. The map says A goes to B. It doesn't say whether B is as good as A was. In most migrations that bleed despite the 301s, it wasn't.
The measurement that wasn't ready
Then there's the part that turns a small problem into a disaster: migrating blind. Analytics and Search Console not set up and validated before launch. The result: nobody sees the bleeding when it starts. You see it weeks later, when it costs twice as much to recover and half the positions have cooled.
The map tells you which URLs should redirect. It doesn't tell you what's actually happening after you hit the button. Measurement ready before the switch is what separates a problem caught the same day from one that only shows up in next month's report. Without it, the migration doesn't have a dashboard. It has a rear-view mirror. The Search Console report and your attribution models get wired up before, not after.
The absence at launch
And finally, the detail. A migration sinks in what nobody checks at go-live: a wrong response header, a canonical pointing to the wrong place, a robots line blocking what it shouldn't, a sitemap nobody submitted. None of this is a map failure. It's a presence failure.
I take few migration projects at a time, and it's not a pose. It's that a migration is won or lost at launch and in the days right after, with someone watching the numbers in real time. Not three weeks before, in the spreadsheet. Preparation prevents the disaster. Presence catches what preparation didn't foresee.
Verdict: do the map, then do what the map doesn't
Make no mistake: do the redirect map, 1 to 1, flawless. It's necessary. But a perfect map on top of ungoverned decisions, gutted content and an Analytics nobody switched on bleeds anyway. The migrations that keep their traffic aren't the ones with the best spreadsheet. They're the ones where someone owned the decisions, protected the signals, had measurement ready and was there at launch. That someone is the job. It's what I do when I take on a migration.
Treat a migration as a redirect exercise, and you migrate the URL and lose the business it brought.
Sources
Frequently asked questions
If I do 1:1 redirects, am I safe in a migration?
No. The map is necessary, not sufficient. The losses almost always come from what the map doesn't cover: decisions made outside the plan, content and internal-link signals a 301 doesn't carry, and measurement that wasn't ready to catch the drop in time. See SEO Migration →
I did everything right and lost traffic anyway. Why?
In most cases, the redirect destinations were weaker pages than the originals, or the new architecture scattered the internal links that fed those pages. The URL moved, the signal stayed behind. It's worth auditing destination by destination, not just confirming the 301 responds. Technical SEO checklist →
Which redirect should I use in a migration?
301 (permanent), always, no chains or loops. Google recommends permanent redirects, says they don't cause a loss in PageRank, and advises keeping them for at least a year, the time it takes to transfer all signals to the new URLs.
How long does it take to recover after a migration?
It depends on how early you catch the problem. With measurement ready and presence at launch, days to a few weeks. Blind, months, and not always in full.