Moving to a new web host with zero downtime

Think of a migration as three separate acts: copying the site, proving the copy, then redirecting traffic. Your old host neither knows nor cares that a duplicate exists somewhere, and keeps serving your domain exactly as it did yesterday. All the risk lives in act three, the DNS switch, and that is precisely why it goes last, once the copy has been shown to work.

In practice that means keeping the old hosting paid up while the move happens, because an overlap sets you back a few dollars and an outage sets you back rather more than that. Carry everything across, test it thoroughly on the new server, and only after that point the domain. Cancel the old plan a week or two later, once nothing is still looking back at it.

First mistake: flipping DNS before the copy has been proved. Visitors turn up at a half-finished site and you end up debugging in production while customers watch. Every migration-downtime story that reaches us starts right there, and the copy-first rule exists to head it off.

Second mistake: closing the old account too early. Plenty of providers delete everything the moment you cancel, so if something on the new copy turns out broken, the working original has already gone. That old account is your rollback. Hang on to it until the new one has carried genuine traffic for several days.

A word about lock-in. Hosts that make leaving awkward, with no backup exports, a retention gauntlet waiting at cancellation, or transfer-out fees stuck on domains, are telling you plainly how they hold on to customers. Check portability before the day you need it. That is also our reason for running migration as a free, verified, zero-downtime service: simple to join, with nothing technical standing in your way should you ever decide to leave.