Mistakes and fixes
Site migration backlinks: list the URLs worth a 301
Redirect maps usually start from a crawl. Start from your links instead. The URLs people have linked to are the ones you cannot afford to get wrong.
Why links should drive the redirect map
A crawl finds pages that exist. It does not find the pages the web remembers, such as old campaign URLs, deleted PDFs and a retired blog category. These may have no internal links left, yet they keep collecting visits from sites that mention you. Easy to miss.
That is why the first export in any move is the list of linked URLs. Call toppages on the old domain, add the Search Console Pages export, and merge the two, keeping every URL that appears in either list. Nothing gets dropped.
Cost is trivial. One call is 2 credits, or $0.0002, so even a site with thousands of linked URLs needs only a handful of calls. You can pull a fresh list the day before launch to be sure nothing slipped through.
The merge formulas are on the join page. Reuse them, swapping in the old domain, and the result is a single list sorted by what the web linked to and what Google actually served. Quick.
Seven mistakes we would look for first
| Mistake | What it causes | Fix |
|---|---|---|
| Redirect everything to the homepage | Google can treat these as soft 404s, and the specific page's signal is lost | Map each old URL to its closest new equivalent |
| Chains of 301s over several hops | Slower loads and extra points of failure | Point every old URL straight at the final address |
| Blocking old URLs in robots.txt | Crawlers cannot follow the redirect | Allow them until the move has settled |
| Switching www and non-www midway | Two hosts split reports and links | Decide once, then 301 the other host |
| hreflang pointing at redirected URLs | Alternates fail to confirm each other | Update every hreflang to the new canonical URLs |
| Dropping redirects after a few weeks | Old links go dead again | Keep them for at least a year, as Google's migration guidance suggests |
| Not re-checking Search Console | Errors sit unseen | Add the new property and watch coverage daily for the first weeks |
A runbook for the week before the move
Export linked URLs
Run toppages for the old domain and download the Pages report from Search Console for the last 16 months, the longest window it keeps.
Merge and rank
Combine both lists in Sheets and sort by referring domains, then by clicks. The top 200 rows are your priority set.
Write the map
Give each priority URL one destination. Where nothing matches, choose a parent category rather than the homepage.
Test before launch
Request every mapped URL on a staging copy and check for a single 301 and a 200 at the end.
Compare after launch
Next month, pull toppages again for the new host and diff it with the old list. Gaps show where redirects broke.
What to do with the URLs nobody links to
Not every old URL deserves a hand-written redirect. Pages with no links, no clicks and no meaning can return a 404 or 410 on purpose, and Google copes with a deliberate removal better than with a redirect that makes no sense. Be decisive.
Set a cut-off early. In our example sheets the line sits at no referring domains and under 10 clicks in 16 months, although your own site may need a different bar. Write the rule at the top of the map so a colleague can check it.
Remember that redirects pass on what the web already did, and no more. If the new page is thinner than the old one, the links will help less than you hope. Treat the move as a content check as well, and review the Red band of the triage table once the dust settles. Short on time? Do that first.
Realistic expectations
A clean move limits the damage, nothing more
Traffic often dips after a migration, even a careful one, and no redirect map guarantees the old positions will return. What a link-driven map gives you is a shorter list of things that could have gone wrong, which is the thing you need at 2 a.m.
- Our data is monthly, so run the export early
- Search Console remains the source for your own live errors