Getting Ahead of Shortening Certificate Lifetimes

Certificate Lifetimes Are Collapsing, and Manual Renewal Is Running Out of Road

Most IT admins have lived this story at least once. The production service goes down. The website won’t load, the API returns errors, and something business-critical is broken. Nobody deployed anything, nothing changed, and the person who stood up the service last year is already at their next job. Fifteen minutes into the incident, someone checks the certificate and finds it expired yesterday. The fix is quick, the new certificate is good for another year, and everyone agrees the process will be better by then.

Surely.

I have watched that exact incident play out at multiple organizations, often more than once at the same organization with the same certificate. It is embarrassing, it is entirely preventable, and it is about to get significantly harder to avoid.

What Actually Changed

In April 2025, the CA/Browser Forum (the body that sets the baseline requirements every publicly trusted Certificate Authority must follow) passed ballot SC-081v3. The vote was 29 in favor and zero opposed, with five CAs abstaining. Every browser vendor with a root program voted yes.

The ballot reduces two things on parallel tracks; one for the maximum lifetime for a TLS certificate, and one for how often organizations must prove they manage the domain on the cert. These values are noted as Max TLS Certificate Lifetime and Max Domain Validation Reuse.

 

As of the publishing of this article, we have already passed the first milestone. It is already the case that Public CAs will no longer issue a TLS certificate valid for more than 200 days.

Two details are worth understanding clearly to prevent confusion.

First, the cap applies at issuance, not retroactively. Nothing you already hold stops working on a milestone date. A certificate issued on March 14, 2027, with a 100-day validity remains valid for its full term. What changes is that after each date, you cannot obtain anything longer.

Second, many CAs deliberately shave a day off the maximum. Your 200-day certificate may be good for only 199, and your 100-day certificate for only 99. Renewing on the day of expiry is a bad strategy regardless, but it is worth knowing that the number on the box is the max and not necessarily what you will receive.

Putting Numbers On It

Venafi says the average enterprise deals with roughly 3,370 publicly trusted certificates today. They also expect that number to approach 5,000 by 2027. For the sake of argument, let’s round the number of certificates that have to be managed down to 3,000. Certificate count is not what changes; renewal frequency is.

 

Today, 3,000 certificates means about 2,700 renewals a year, roughly ten on a working day. That is genuinely manageable with a spreadsheet, some calendar reminders, and expiry emails from your CA. It is ugly, but it works, which is exactly why this approach is so common. Many organizations push it further and make each application team responsible for its own expiring certificates, with the PKI or IAM team acting as an order desk. That distributes the work, but it also distributes the failure.

At 47 days, the same 3,000 certificates generate close to 64 renewals every single day, including weekends and holidays. If you assume 15 minutes per certificate end to end for discovery, request, validation, installation, and verification, that is two full-time people doing nothing else. Fifteen minutes is also wildly optimistic for many issuances. Some renewals take two days because the certificate lives on an appliance that needs a maintenance window, or because you are waiting on a change ticket and CAB approval.

Sixty-four renewals a day is not a staffing problem you can hire your way out of. It is a process that must stop involving humans.

Why This Is Happening

Partner with Microsoft experts you can trust

If it’s time to take that first step toward leveling up your organization’s security, get in touch with Ravenswood to start the conversation. 

It is worth understanding the reasoning because it changes how you prioritize the work.

Revocation often does not function. If a key is compromised, revoking the certificate only helps if clients check revocation, and in practice many do not. A three-year certificate is trusted for three years regardless of what you publish to a CRL if clients do not check it. Short lifetimes are a blunt but effective substitute: the certificate simply stops being valid on its own.

Certificates stop outliving domain ownership. A certificate’s validity period and a domain’s validation reuse period are separate timers, and conflating them understates how long stale control can linger. Domain control validation only gates issuance. The real worst case is additive: a certificate can be issued right at the edge of the reuse window using validation that’s about to expire, then run its full validity period from there. The reuse period shrinking to 10 days doesn’t mean stale control clears in 10 days; it means whatever gets issued in that window still gets its own 47 days to run.

Below seven days, revocation infrastructure stops mattering as much. Certificates valid for seven days or less are exempt from CRL and OCSP requirements entirely, because the validity window is shorter than the useful life of a revocation response. Let’s Encrypt already offers six-day certificates. For anyone who has spent a weekend explaining how a broken CRL or firewall rule took down an environment, that is a genuinely appealing future.

Your Internal PKI Is Exempt, and That Might Be a Problem

Everything above applies to exactly one category of certificate: publicly trusted TLS. Baseline Requirements govern those and nothing else. Your internal ADCS issuing CA can hand out ten-year certificates tomorrow, and no browser will object.

That exemption is not a reprieve. It is the reason internal PKI is the part nobody is coming to fix. Public certificates are getting solved because the browsers are forcing the issue. Internal certificates have the same silent failure mode with none of the external pressure, which is how organizations end up with critical services that rely on an internal CA quietly counting down toward an outage that will take out authentication for the entire company.

Any of the many services Active Directory (AD) autoenrollment does not cover is a calendar-reminder-shaped outage waiting to become technical debt.

A growing number of tools provide support Automated Certificate Management Environment (ACME) frontends and shims for Active Directory Certificate Services, (AD CS) and it is not hard to imagine a future where ACME is the universal issuance protocol and the only thing left to solve per-endpoint is deployment.

What Admins Should Actually Do

  1. Build an inventory you don’t have to ask anyone for. You cannot automate what you cannot see. Certificate Transparency logs are free through services like Censys and will show you every publicly trusted certificate issued for your domains, including the ones a team ordered without telling you. Pair that with active network scanning to catch what is in use. 
  2. Get off email and manual file-based domain validation now. Move every domain you use to automated DNS-01 validation. This usually means delegating the _acme-challenge records (often via CNAME to a dedicated validation zone) and making sure your DNS provider has an API your tooling supports. If your DNS is manually managed, that is the first real blocker to clear, and it is frequently owned by a different team.
  3. Stand up ACME issuance. This is the easy win and there is mature tooling for it. Certify the Web, win-acme, and certbot cover most on-premises scenarios, and every major cloud and CDN offers managed certificates. Most commercial CAs now expose ACME endpoints, so you can keep your existing CA relationship if you have one.
  4. Triage your endpoints by how hard they are to automate. Sort every certificate consumer into three buckets: fully automatable today, automatable with scripting, and requires a human. That third bucket is your real project plan. Appliances with GUI-only certificate upload, Java keystores, and vendor middleware belong there. For each one, decide now whether you will script it, replace it, or move the TLS termination in front of it to something you can automate. Terminating at a load balancer or reverse proxy you control is very often the cheapest answer.
  5. Fix change control before it becomes the bottleneck. A process requiring CAB approval per certificate deployment does not survive 64 renewals a day. Get automated certificate rotation classified as a standard or pre-approved change while it is still a planning exercise rather than an incident.
  6. Monitor the endpoint, not the pipeline. Alert on the certificate a service actually presents, with thresholds tied to a percentage of validity rather than a fixed number of days. A 30-day warning is meaningless on a 47-day certificate. Assume renewals will fail silently and design so that you find out in hours, not on the morning of expiry.
  7. Give your internal PKI a deadline of its own. Nobody will impose one on you. Inventory your internal certificates, find the ones autoenrollment does not cover, and shorten their lifetimes deliberately. A certificate you rotate every 90 days by automation is a certificate you know still rotates. Long internal lifetimes are not safety; they are just a longer interval between tests of a process nobody has verified.

Conclusion

The best time to automate certificate management was several years ago. The second-best time is now, while 200-day certificates still leave enough slack to build something carefully. Organizations that wait will spend the next two years doing an enormous amount of manual work under increasing time pressure and will end up automating anyway.

Are you looking for help with automating certificate issuance? Contact an RTG expert today!

[RELEVANT BLOG CONTENT]