Design Status Pages and Uptime Claims Honestly
Feedback & States · status, trust, errors
Updated 2026-08-04
- Know what your uptime number actually permits. Three nines allows almost nine hours of downtime a year, and four nines allows under an hour, so the figure on the marketing page is a promise with a specific size.
- Host the status page somewhere that survives your outage. A status page served by the infrastructure that is down is the most reliably useless page on the internet.
- Say what is degraded rather than whether the system is up. Products fail in parts, and a single green light is wrong more often than it is right.
- Update during the incident, not after it. A page that says operational while users cannot sign in costs more trust than the outage does.
- Give each incident the four things people need:
- What is affected, in product terms rather than service names.
- When it started.
- What is being done.
- When the next update will come.
- Link to it from the product at the moment of failure. Most users never find a status page, and an error message is the one place they are already looking.
- Publish partial degradation as its own state. Slow is not down, and users planning their day need to know which one they are in.
- Keep history visible. A published record of past incidents is a stronger trust signal than an unbroken run of green, because nobody believes the unbroken run.
An uptime figure is a claim, and the status page is where a product proves it means it.
Related guides