Expired certificate outages are almost a genre of their own in postmortems — big, well-resourced companies included. The failure is rarely a lack of knowledge. Every engineer knows certificates expire. The failure is almost always process: nobody owned the renewal, or the alerting didn't reach the right person in time.
What actually happens when a certificate expires
Browsers and clients that connect over HTTPS validate the certificate presented by the server. Once it's past its expiration date, that validation fails — visitors see a security warning, and depending on the client, many API integrations and mobile apps will simply refuse the connection outright rather than showing a warning at all. For a customer-facing service, that's a hard outage, not a degraded experience.
Why this keeps happening despite being preventable
- Ownership gaps — a certificate issued during a project has no clear long-term owner once that project ships.
- Short validity periods — many public certificate authorities now issue certificates valid for 90 days or less, which means renewal has to happen far more often than the old one-year cycle teams are used to.
- Alert fatigue — a single reminder email 30 days out is easy to miss if it's not tied to an actual action item with an owner.
- Shadow infrastructure — internal tools, staging environments, and legacy systems often carry certificates nobody's actively tracking at all.
What actually prevents it
The fix isn't "try harder to remember." It's removing the dependency on any one person remembering. That means centralized discovery of every certificate in your environment, automated renewal wherever your certificate authority supports it, and alerting that escalates — not a single email that's easy to archive and forget.
The bigger pattern
Certificate expiry is a small, specific example of a much bigger theme in IT management: tools and systems that were set up correctly once, but have no ongoing owner. It's the same pattern behind forgotten software licenses, stale user accounts, and abandoned infrastructure — the initial setup was fine, the maintenance was where it broke down.