Why am I getting 404 errors in Search Console after a redesign?

Those 404 errors in Search Console after a redesign have a single common cause: the redesign changed your URLs, and the old ones were never redirected. Google still has the old addresses indexed, so it keeps requesting them, and now they return nothing. The report is just Google telling you what broke. The fix is direct: export the 404 list, map each old URL to its new page, and add a 301. WPBuildAI crawls the old URLs and builds that redirect map, so the 404s do not appear in the first place.

Search Console gained a new place to look in 2026 as well: Google launched Search Generative AI performance reports in June 2026, so alongside chasing down 404s you can now see whether the redesigned URLs are still surfacing inside Google’s generative answers.

What the 404 report is really telling you

A 404 in Search Console is a URL Google knew about that no longer resolves, and reading the report correctly changes how you treat it. Google does not invent these URLs; they are addresses it previously crawled and indexed, often pages that ranked and earned links, that now return not-found. So the report is a list of things that used to work and stopped, which is exactly your recovery list. The harm is not the error message itself but what it implies: each of those URLs lost its rankings and any backlinks pointing to it the moment it stopped resolving. Ahrefs found in its search traffic study that a small set of pages carries most organic traffic, so if the 404 list includes some of those, the traffic loss is real and measurable. Treat the report as a prioritised recovery list, not just a warning to dismiss.

A hard 404 versus a soft 404

It helps to distinguish the two, because they look different in the report and have the same fix but a different cause. A hard 404, the subject here, is an old URL that genuinely returns a not-found status: Google requests it and the server says “gone,” which is honest but means the page and its rankings are unreachable. A soft 404 is different: the URL returns a success status (200) but shows no real content, so Google concludes it is effectively missing despite the server claiming it is fine, the situation covered in soft 404s after a redesign. Both appear in the coverage report and both come from a redesign mishandling old URLs, but the hard 404 is an honest dead end while the soft 404 is a misleading one. The fix for both is to map the old URL to a real, relevant destination with a 301, or to return a deliberate 404/410 for genuinely retired pages.

Clearing the list correctly

Export the 404s from Search Console and handle each one deliberately:

• Redirect the URL with a 301 to its closest relevant new page, per Google’s site move guidance and redirects guidance, which confirm the 301 carries the ranking signal to the new URL.
• Return a real 404 or 410 for pages you genuinely retired with no equivalent, so Google drops them cleanly rather than retrying them indefinitely.
• Never blanket-redirect to the homepage, which Google treats as a soft 404 and which passes no equity.

The 2024 Web Almanac shows how many URLs a real site accumulates, which is why whole groups, archives, tags, paginated pages, get missed in a redesign and show up here. These are the same breaks as the internal side of fixing broken links, approached from the search engine’s view.

A worked example: a coverage report full of 404s

Picture a site three weeks post-redesign whose Search Console coverage report shows 180 not-found URLs, alarming until you work through them. Exported and sorted, they fall into groups: 60 are old blog posts whose slugs changed, each with a clear new equivalent, so they get 301s to the matching new posts. 40 are paginated archive URLs (/page/2/ and so on) that the new structure handles differently, redirected to the relevant section. 50 are old tag pages, some worth redirecting to a category, others genuinely retired and given a clean 410. 30 are campaign URLs that are truly gone with no successor, also 410’d. Within a day the list is handled: most as 301s recovering rankings, the rest as honest 410s telling Google to drop them. Over the following weeks the coverage report clears as Google recrawls. The 180 errors looked like a crisis and were really a sorted, fixable worklist.

Why not the homepage

The tempting shortcut is to redirect every 404 to the homepage in one rule, and it is a trap that recreates the problem. Google treats an irrelevant blanket redirect to the homepage as a soft 404, because a user who wanted a specific page does not have their request satisfied by landing on the front page, so the redirect does not actually resolve anything. As a result it passes little or no equity, and the URL effectively stays “missing” in Google’s eyes despite the redirect. So the homepage dump fails on both counts: it does not help the user and it does not recover the ranking. Map each 404 to its closest relevant new page instead, a successor post, a parent category, a related product, and reserve a true 404/410 for the genuinely retired. The homepage redirect feels efficient but converts a hard 404 into a soft one, which is no improvement.

Which 404s to fix first

Not every 404 carries equal weight, so prioritise the list rather than working through it blindly. Sort the exported 404s by the traffic and links the URLs used to have, using your analytics, the URL’s former rankings, and any backlink data. The pages that drove visits or hold backlinks are where a missing redirect costs real, recoverable rankings, so they get fixed first with precise 301s to their true replacements. The long tail, URLs that had little traffic and no links, still gets handled, but a quick rule-based redirect or an honest 410 is fine for them. This matters because Ahrefs’ finding about traffic concentration means a handful of the 180 URLs in the example probably account for most of the lost traffic, so getting those few exactly right recovers most of the damage. Prioritising turns a daunting list into a high-leverage first pass followed by cleanup.

Preventing the next round

The 404 report is the symptom of a skipped redirect map, so the prevention is to never skip it. Before the next redesign or migration, inventory every URL, by crawling the old site and reconciling against analytics, Search Console, and backlink data, and map each to a 301 (or a deliberate 410) before launch, using a URL mapping template. Done that way, the new site goes live with every old URL already pointing somewhere real, so Google never finds a not-found page and the coverage report never fills with 404s. The post-launch cleanup described here is what you do when that inventory was skipped; the inputs are identical, a complete URL list and a redirect map, just applied after the damage instead of before. Unaddressed, the 404s compound into the broader drop described in why traffic falls after a redesign, which is the reason to both clear them now and prevent them next time.

Common mistakes clearing Search Console 404s

The recurring errors slow recovery or waste it. Ignoring the 404 report as “just errors” leaves recoverable rankings stranded. Blanket-redirecting everything to the homepage converts hard 404s into soft ones and passes no equity. Fixing the URLs in arbitrary order, rather than by traffic and links, spends effort on pages that did not matter while the valuable ones stay broken. Redirecting genuinely dead pages to vaguely related ones, instead of returning an honest 410, muddies the index. And clearing the current list without preventing the cause means the next redesign refills the report. Each is avoided by treating the report as a prioritised recovery list: export it, sort by value, 301 each to its true replacement, 410 the genuinely retired, never dump to the homepage, and inventory URLs before the next launch so the 404s never appear.

Key points to remember

404 errors in Search Console after a redesign mean old indexed URLs changed without redirects, so Google still requests them and finds nothing, which strands the rankings and links those URLs held. Export the list and handle each deliberately: 301 to the closest relevant new page for anything with an equivalent, a clean 404 or 410 for genuinely retired pages, and never a blanket redirect to the homepage, which Google treats as a soft 404. Prioritise by former traffic and links so the valuable URLs are recovered first, and prevent the next round by inventorying every URL and mapping redirects before launch. WPBuildAI crawls the old URLs and builds the complete redirect map before launch, so the new site goes live without stranded URLs; send your site URL for a fixed quote.

Not affiliated with Google, WordPress, or Lovable.