The relaunch did not lose your traffic. The missing redirect map did.

The pattern is consistent enough to predict. A company relaunches a site that badly needed relaunching. Two weeks later, organic traffic is down by a third and nobody can explain it, because everything about the new site is better: it is faster, it works on a phone, the copy is not from 2018. The theory that forms in the room is that the new content is not ranking yet and it needs time.

Sometimes that is true. More often, the old addresses stopped answering. A page that spent six years accumulating links and search history is not content, it is an address that content lived at, and the two are separable in exactly the way that costs you.

What a redirect map actually is¶

A redirect map is a list with two columns: every address the old site answered, and the address on the new site that now serves that purpose. Google's own migration guidance names this as the first task, describing it as creating a mapping of old to new URLs and deciding where each old URL should point. The guidance is explicit (opens in a new tab) that this is a thing you build, and the reason it has to be built rather than derived is that no algorithm knows your old site's intent.

The hard half is the left column. The set of addresses that matter is not the set somebody designed; it is the set that ever earned a visit, and it includes the pages nobody remembers: the old press release that a trade publication linked to, the PDF somebody cited, the product page for a line you stopped selling, the category page an old campaign pointed at. Building the left column from the new site's sitemap finds none of them, which is why the map has to start from evidence.

Permanent and temporary are not a style choice¶

A redirect is either permanent or temporary, and the difference is a statement about intent rather than a style choice. Google's documentation (opens in a new tab) puts it plainly: a permanent redirect shows the new target in search results, and a temporary redirect shows the source page. During a move you want the first. Shipping the second, which is a common default in server configuration and in more than one content platform, means the results keep pointing at addresses you intend to retire.

How long to keep them¶

Longer than feels necessary. Google's advice is to keep the redirects for as long as possible, generally at least a year (opens in a new tab), so that all signals can be transferred to the new addresses, which requires recrawling that happens on the crawler's schedule. The same guidance sets expectations on the move itself: a small to medium site takes a few weeks for most pages to move, and visibility fluctuating during that window is normal.

That last point is worth holding onto, because it cuts both ways. Some of the drop after a relaunch genuinely is the settling period and recovers on its own. The way to tell the difference is not to wait and see: check whether the old addresses answer. If they do, you are in the settling period. If they do not, waiting will not fix it and the interval is costing you.

Doing it in the order that works¶

1. 1.Export the old addresses before the old site is switched off. This is the only step with a deadline attached, and it is the one that gets missed, because the old site is usually decommissioned the same week.
2. 2.Build the map as a file, in version control, with a decision on every row. A spreadsheet somebody has is not a deliverable; the map is part of the build.
3. 3.Implement the redirects as permanent ones, and test a sample of them against the live site rather than reading the configuration.
4. 4.Keep a list of what deliberately 404s, so that the dead ends are a decision somebody made rather than an omission somebody will rediscover.
5. 5.Watch Search Console for the addresses you missed, and keep adding rows. The map is never quite complete on the first pass and that is expected rather than a failure.

None of this is expensive if it is done during the build, and all of it is expensive afterwards, because the recovery is the identical work performed under pressure while somebody senior asks weekly whether the traffic is back. The map is the cheapest insurance available on a relaunch, and it is the item most likely to be cut when the launch date is in trouble.