404 Redirect Candidate Finder

Every crawl produces a list of 404s, and on most sites that list runs into the hundreds. Faced with it, teams do one of two things: redirect everything to the homepage, or nothing at all. Both are wrong, and the second is usually less damaging than the first.

The problem is that the list treats every broken URL as equivalent when they are not remotely. A discontinued product with fifteen external links and residual search traffic is a business problem. A malformed URL from a broken link on someone's forum post in 2019 is noise.

Prioritising by inbound links and residual traffic turns an unusable list into a short, high-value one.

A 404 receiving external links is discarding authority someone gave you. Those links still exist and still carry value; they are just pointing at nothing. Redirecting to the closest equivalent recovers most of it, and this is usually the single highest-return action in a technical audit.

URLs still receiving visits, from bookmarks, old emails, printed material or search results that have not updated. Each visit is a person hitting an error page. Check analytics or server logs, not just the crawl.

Your own pages linking to 404s. Entirely self-inflicted and entirely fixable. The correct fix is to update the link, not to redirect the broken URL — a redirect masks the problem and adds an unnecessary hop.

No links, no traffic, no history. Malformed URLs, spam requests, long-deleted test pages. These need no action. Leaving a 404 as a 404 is the correct response when the page genuinely does not exist and nobody is looking for it.

The test for a redirect target is whether a person who wanted the old page would be satisfied by the new one.

• Discontinued product → the closest current alternative, or the parent collection if there is no equivalent.
• Deleted article → the article that replaced it, or the most relevant related piece.
• Removed collection → the parent category or the collection that absorbed it.
• Changed URL structure → the direct equivalent under the new structure.
• Genuinely gone with no equivalent → leave it as 404, or return 410 to signal permanence.

410 Gone tells search engines the page is deliberately removed and will not return. It gets processed and dropped faster than a 404, which lingers in case it was temporary.

Use it for pages you have intentionally deleted with no successor — expired campaigns, discontinued lines with no replacement, removed user content. It is a small signal, but it keeps crawl budget focused on pages that exist.

• Redirect at deletion time. Set up the redirect in the same action that removes the page, rather than as a later cleanup task.
• Crawl before and after migrations and compare. Most large 404 clusters originate in a migration.
• Watch product lifecycle. On ecommerce sites, discontinued products are the largest recurring source, and they are predictable.
• Check internal links after restructures. Renaming a collection changes its URL and breaks every link pointing at the old one.
• Monitor Search Console coverage monthly rather than waiting for a full audit.

A few platform-specific patterns account for most Shopify 404s.

• Deleting a product creates a 404 with no redirect. Shopify does not generate one automatically, so discontinued products with inbound links quietly lose their value.
• Renaming a collection changes its handle and therefore its URL. The old URL 404s immediately.
• Unpublishing versus deleting behave differently. Unpublished products return 404 while still existing in the admin, which confuses audits.
• Bulk imports overwriting handles can change URLs silently across many products at once.

Redirects are managed under Navigation in the admin and support CSV import, which makes bulk fixes practical once you have a prioritised list.

The prioritisation depends on data a crawl alone cannot give you, so the useful version of this audit combines three sources.

1. Crawl data for the list of broken URLs and which internal pages link to them.
2. Backlink data for which broken URLs still receive external links, and how many.
3. Analytics or server logs for which broken URLs are still being requested by real visitors.

Any URL appearing in the second or third list is worth a redirect. URLs appearing only in the first are internal link fixes. URLs in none of them can be left alone.

If you do not have backlink data to hand, Search Console's Links report shows externally linked pages, and the coverage report flags 404s Google has encountered. Between them you can identify most of the URLs that matter without additional tooling.

A soft 404 is a page that returns a 200 status while containing nothing useful — an empty search results page, a collection with no products, or a redirect to somewhere irrelevant.

These are harder to find than genuine 404s precisely because the status code says everything is fine. Google identifies them anyway and treats them as errors, but your crawl will report them as healthy pages.

Two common sources on ecommerce sites: collections that have emptied as products sold out, and search or filter pages that return no results but still render a full template. Both look normal in a crawl export and both need the status code corrected or the page removed.