How to set up a public status page for your SaaS

When your app breaks, customers first ask: is it me or is it you? A public status page answers that before they open a ticket. This guide shows what to put on the page, where to host it, who updates it and what to write.

  • Lalida, UptimeVanguard

What a status page is for

A status page is a public web page that shows if your service works right now. It also shows the problems you know about and what you do about them.

It tells customers about a problem before they report it, and it cuts the number of support tickets during an outage.

Show components your users recognize

List the parts of your product that customers use, in their words: "Web app", "API", "Login", "Payments", "Email delivery". Do not list servers, databases or queue workers. A customer does not know what "db-02" is.

Keep the list short. Each component needs a check behind it, so its status changes without a person. A component that is always green because nobody updates it does more harm than no component at all.

If your product depends on outside services, such as a payment provider or a cloud host, you can show their status too. Then customers can see when the cause is outside your control.

Host it away from your own infrastructure

The status page must work when your product does not. If it runs on the same servers, the same network or the same cloud account as your app, one outage takes both down.

Use a hosted status page service, or at least a separate provider. Test it: when your app is down, can you still post an update?

Decide who updates it, and how fast

Automatic checks show that something is down. They do not say what happened or when it will be fixed. A person must do that, so decide before the next incident:

  • Who posts. The person on call, or a named role in the incident. Give more than one person access, so a holiday does not block an update.
  • How fast. Post the first update within minutes of a confirmed problem, even if you know little. "We are investigating" is better than silence.
  • How often. Post again at a fixed interval during the incident, for example every 30 minutes, even when nothing changed. Say when the next update comes.

What to write in an update

Write for a customer, not for your engineers. A good update answers four questions:

  • What is affected? "Some card payments fail at checkout."
  • Who is affected? All users, one region, or one feature.
  • What are you doing? Investigating, fix identified, monitoring.
  • When is the next update? A time, not "soon".

Use the same stages each time: investigating, identified, monitoring, resolved. Do not paste raw error messages or stack traces. After a large incident, post a short summary of the cause.

Let people subscribe

Most customers do not watch your status page. Let them subscribe, so the news comes to them. Email is the minimum. Ask subscribers to confirm their address first, and put an unsubscribe link in every message.

Plan your maintenance and announce it on the page in advance, so subscribers can plan around it.

History builds trust

A page that shows only "All systems operational" looks like marketing. A page with past incidents and daily uptime bars shows that you report problems. Keep resolved incidents visible. Do not delete an incident because it looks bad.

How UptimeVanguard does it

In UptimeVanguard, a status page is part of your monitoring. You choose which monitors appear as components, or all of them. Each component shows the live status of its checks. The page runs on our servers, not yours, at an address like /status/your-page, with a short link at /s/your-page. When a monitor goes down, an incident opens on it and the page shows it without any action from you. The page refreshes itself every 15 seconds.

The page shows:

  • Current status: components with issues, the healthy ones, active incidents and active and scheduled maintenance.
  • Uptime history: a bar for each of the last 90 days per component, with the uptime percentage.
  • Incident history: incidents of the last 30 days with the updates you wrote. Automatic messages with check errors are not shown.
  • Third-party services: an optional block with the status that each outside service you follow reports.

On an incident, you post updates with the stages Investigating, Identified, Monitoring and Resolved. Templates fill in a first text. An update goes out now or at a time you set. You can also post a status update on the page itself, without an incident.

Visitors subscribe by email and confirm by link. They get an email when a component goes down or recovers, and when you post an update, unless you turn that off for the page or for one update. You add your logo and an accent color, and choose auto, light or dark appearance. A switch takes the page offline while you set it up.

The Free plan includes 1 status page. On the paid plan, 5 sites cost €10.00 a month, excl. VAT, with up to 5 status pages per team. To set up the checks behind the components first, read how to get an alert when your website goes down, or start for free.

Questions

Questions people ask

Should a status page show every server?

No. Show the parts of the product that customers recognize, such as the web app, the API and payments. Internal names confuse customers and tell attackers about your setup.

How often does an UptimeVanguard status page refresh?

Every 15 seconds while the page is open. The uptime history bars cover the last 90 days.

Do subscribers get an email for every incident?

Confirmed subscribers get an email when a component on the page goes down and when it recovers. Updates you post are emailed too, unless you turn that off for the page or for one update.

How many status pages can I make?

One on the Free plan. Up to five per team on the paid plan.

Start now

Tell customers before they ask.

Start free: every check type on 5 sites for 7 days, then the Free plan: 3 sites with every check type, every alert channel and 1 status page. No card.