Missing pages return a real 404

Why does your 404 page matter for SEO?

Because what your site says when a page is missing is as important as what it says when one exists. A site that reports missing pages as though they were fine teaches search engines to distrust everything else it reports.

What it actually is

Every response carries a status a visitor never sees: this worked, this moved, this is gone. When someone reaches an address that does not exist, the honest answer is "not found". A soft 404 is a site that shows a friendly missing-page message to the human while telling the machine everything was fine — the visible page and the underlying signal disagreeing.

Why it matters

Crawlers budget their attention. A site that reports success for addresses that do not exist invites them to collect an unlimited number of identical non-pages, spending that budget on nothing while your real pages wait. Worse, those phantom pages can end up indexed, so a search result promises content that was never there. For AI assistants the failure is sharper still: a confident "this page exists" is taken at face value.

What good looks like

A missing address returning an honest not-found signal, paired with a page that actually helps the person who landed there — your branding, a route back to the main sections, and a search or contact option. Honest to the machine and useful to the human, at the same time, on the same response.

Being honest about it

Nobody has ever grown a business on their 404 page and we are not going to claim otherwise. This matters as a correctness signal rather than a growth one, and its real value is diagnostic: a site getting this wrong is usually getting other status-code questions wrong too, in places that cost more.

Where most sites go wrong

By far the most common cause is a platform or redirect rule that quietly sends every unmatched address to the homepage. It feels helpful and it is quietly corrosive: the site now claims infinite copies of the homepage exist at infinite addresses, which is both a duplicate-content problem and a credibility one. The second pattern is a genuinely correct signal attached to a bare, unbranded error page, which is a smaller problem but a worse experience.

← See all 64 checks