How to migrate without losing search traffic

This site used to say that a migration can be done without losing search traffic. That was a sentence I could not stand behind, and it has been rewritten.

Here is the version that holds up. Search equity can be protected. Rankings cannot be promised. Those are two different claims and the gap between them is where most of the disappointment in this work lives.

What carries and what does not

Equity is the accumulated signal attached to your URLs: the links pointing at them, the history, the relevance a search engine has assigned over years. A 301 passes that on. Done properly, the new address inherits what the old one had.

What does not transfer is everything that was never about the URL.

Anyone who guarantees the right-hand column is either not thinking about it or is counting on you not to check. The thing worth buying is the left column done thoroughly, because it is the part that is actually within anybody's control.

The sequence that matters

Order is most of the technique here. The same set of tasks done in the wrong sequence produces a worse outcome.

Before anything moves, write down where you are. The full URL list, the top pages by traffic, the queries earning impressions, the current numbers. This document is not bureaucracy: it is the only way to tell later whether something dropped or whether the way you were counting changed.

Build the redirect map against that list, not against a crawl. The reasoning is in redirect mapping, and it is the difference between protecting the site you have and protecting the site you remember.

Keep the URLs where you can. The cheapest redirect is the one you did not need. A rebuild is a tempting moment to fix a URL structure, and sometimes that is right, but every changed address is a small risk taken for a benefit that should be nameable. "It is tidier" is not one.

Launch, then submit the new sitemap immediately. This is what tells the crawler to come and look. Leave the old sitemap reachable for a while too, so the crawler walks the old addresses and finds the redirects rather than discovering them by accident.

Do not change the content and the platform in the same week. This is the one people skip, and it is the one that makes diagnosis impossible. If the pages are rewritten and the URLs move and the templates change simultaneously, and traffic falls, nothing in the data can tell you which of the three did it. Move the site, wait, then rewrite.

What a normal dip looks like

There is almost always a dip. Saying otherwise is the thing I stopped saying.

A search engine has to recrawl the old URLs, follow the redirects, index the new ones and reassign what it knew. That takes time proportional to how often it visits your site, which for most B2B sites is not daily. During that window the picture is genuinely unsettled.

What normal looks like: a drop in the first week or two, a recovery curve through weeks three to six, and a return to the previous range inside two months. Bigger and older sites take longer, because there is more to recrawl. A site with a few dozen pages may not visibly dip at all.

None of that is an argument for being relaxed about week one. It is an argument for not making an emergency decision in week one, because the shape that means everything is fine and the shape that means something is broken look identical for about ten days.

When to worry

Four signals that this is not the normal curve.

Pages are reported as not found in Search Console. The crawler is hitting an address that has no rule. That is a gap in the map, it is findable, and it is fixable the same day.

Traffic to one template collapses while the rest holds. Not a search problem. Something about that template is broken: the markup, the internal links into it, or a rule that is catching more than it should.

Impressions hold and clicks fall. You are still ranking and the listing has got worse. Almost always the missing structured data, or titles and descriptions that were rewritten during the move without anybody deciding to.

Weeks pass and there is no slope. This is the signal, and it is the absence of recovery rather than the depth of the dip. How long to give it depends on how often your site gets crawled, how much of it moved and how much of the content changed at the same time, so there is no week number that means broken. What you are watching for is a line that is flat rather than climbing, while indexing, the old addresses and the coverage report all look correct.

The failure I see most is none of these. It is a team that panics in week two, changes four things at once, and destroys their own ability to tell what the migration actually did. The measurement discipline is the same discipline as the migration itself.

Search is also only half of what a migration can quietly get wrong. Forms, analytics and structured data fail without any visible symptom, and the migration guides go through the platform-specific parts of all of it.