A website migration is a records job, not a design job

Most of the risk in moving a site has nothing to do with how it looks when it lands.

People budget a migration as though it were a redesign. The design is the easy half. The half that goes wrong is the part nobody can see: the addresses.

Every page you have ever published has an address, and other people's links, Google's index and your own old emails all point at those addresses. Move the site without carrying them across and every one of those links lands on a 404. Traffic you spent years earning disappears in an afternoon, and it does not come back on its own.

So the first artefact of a migration is not a mockup. It is a list: every URL on the old site, what it becomes on the new one, and a redirect connecting the two. Where a page is genuinely gone, it redirects to the nearest thing that still answers the question, not to the home page - a home-page redirect tells a visitor you did not care enough to find out what they wanted.

The second is the content inventory. Sites accumulate pages that nobody has looked at in three years, and a migration is the one natural moment to decide what deserves to come. That decision should be made from data, not memory.

Third is the things that are not pages at all: forms and where they send, tracking that has to keep counting, the email that goes out when somebody buys, the feeds other systems read. These are the parts discovered after launch, by a customer, if they are not written down before.