How to redesign your site without losing rankings
Most redesigns that lose traffic skipped the same step, and it happens before anyone touches the design. Here is the full procedure.
There is a story that repeats itself: a company redesigns its site, it looks far better, and three weeks later traffic has halved. The design was not to blame. What happened is that a new site launched without anyone knowing what was holding the old one up.
A site that has been online for years accumulates something no screen shows: URLs Google already knows, pages that receive visits from specific searches, and links from other sites pointing at concrete addresses. If those addresses disappear without notice, all of it goes with them. The good news is that it is entirely avoidable, and the work that avoids it happens before anyone draws a single screen.
Step 1: the inventory almost nobody does
Before touching anything you have to answer three questions about the current site, with data rather than from memory.
- Which URLs receive traffic today. It comes from Google Analytics and Search Console. You will find pages nobody remembered that bring visits every month.
- Which searches you appear for. Search Console gives you the real queries people arrive through. Those queries are what the new site has to keep covering.
- Which external sites link to you and to exactly which pages. If someone linked your 2019 case study and that URL stops existing, the link is lost and part of your authority with it.
Step 2: decide what survives
With the inventory in hand, every URL falls into one of three categories, and you have to decide explicitly which one.
The ones that keep the same address are the best: they need nothing. The ones that change address need a 301 redirect to their equivalent on the new site. And the ones that simply disappear —obsolete content that no longer applies— need a conscious decision: either redirect them to the closest page by topic, or let them die with an honest 404.
The frequent mistake is redirecting everything to the homepage. Google treats that as a soft error: the redirect does not take the user to what they were looking for, so it transfers no value. A redirect only works when it points to equivalent content.
Step 3: the 301 redirect plan
A 301 redirect tells Google “this page moved permanently to this other one”. It is the mechanism that transfers ranking from the old address to the new one, and it is what keeps a structural change from costing you months of traffic.
The plan is literally a table: old URL, new URL. It gets built during the project, not on launch day, and it gets tested on staging before the site goes live. Two details that are frequently overlooked: avoid redirect chains (A going to B and B to C instead of A straight to C), and verify the redirect is a permanent 301 and not a temporary 302, because the temporary one transfers no value.
Step 4: the technical checks before launch
This is the minimum pre-launch checklist. Any of these getting through wrong costs traffic.
- That the new site does not launch with a
noindexinherited from the staging environment. It is the most common mistake and the most expensive. - That the production
robots.txtis not blocking the entire site. - That every page has a canonical pointing at itself and not at the staging version.
- That titles and descriptions did not get lost in the migration.
- That the XML sitemap is updated with the new URLs and not the old ones.
- If the site is bilingual, that hreflang tags remain reciprocal between versions.
Step 5: monitor afterward, not only before
Launch is not the end. Over the first weeks Google re-crawls the site and that is when the problems testing did not catch show up: a missing redirect, a page left out of the sitemap, a 404 on a URL that did have traffic.
What to watch in Search Console is coverage —which pages ended up indexed and which did not— and per-query performance compared against the previous month. A small dip in the first weeks is normal while Google reprocesses. A dip that has not recovered by week eight is no longer explained by reprocessing, and that is when you go find what broke.
How long recovery takes
With the work done properly, traffic usually takes up to eight weeks to stabilize. How long exactly depends on how often Google recrawls your site, which nobody controls: sites that update frequently get reprocessed sooner than sites that have sat unchanged for years. From there the new site tends to outperform the old one, because it is normally faster and better structured.
Without the upfront inventory, recovery can take months and is sometimes incomplete, because external links pointing at dead URLs do not come back on their own. Which is why the step most often skipped is precisely the one that costs most to skip.