Guides

How to build a status page — step by step, so it doesn't go down with your site

Illustration: a small notice board standing on its own, away from the building

The site is down. People go to find out what happened — and hit the exact same error, because the status page lives on the same server.

That's the central trap of a status page: its value appears precisely when everything else is unreachable. Which means there's only one real requirement — it must not depend on the thing it watches.

Let's do it in an evening.

Step 1. Decide where it lives

One rule: not where your app lives.

If the app is on your server, the status page belongs on someone else's. Any free static host works, as long as it's different infrastructure, a different account and preferably a different provider.

A word on the domain. status.yourproject.com looks tidy, but if the domain's DNS goes down, the status page goes with it. Strictly speaking, keep a backup address on another domain. Realistically: use the subdomain, delegate it separately, and know about the risk.

Step 2. Take a ready-made service or build static HTML

Two paths, both fine.

A ready-made service. There are plenty of status page tools, and most have a free tier for one project. Upside: subscriber notifications and incident history out of the box.

Your own static page. One HTML file, no backend: a heading, a list of components ("Website", "API", "Bot"), each with a coloured dot and a line of text. Publish it on static hosting, update it by editing the file.

The second option sounds primitive — which is exactly why it never falls over. No database, no rendering, served from a CDN. For a side project that's enough.

Step 3. Add an automated check

You need an external watchman that knocks on your site every minute from someone else's machine.

  1. Sign up for a free uptime checker (there are dozens).
  2. Point it not at your homepage but at a dedicated check endpoint — say /healthz, which answers 200 and computes nothing.
  3. Set up a notification to your email or Telegram.
  4. Add a threshold: alert after two or three failures in a row, not the first one. Otherwise every network hiccup wakes you up.

The key part: the check runs from outside. A script on your own server won't notice that your own server is unreachable.

Step 4. Write the messages in advance

During an incident there's no time to find the words — your hands shake and you produce either "everything is broken!!" or bureaucratic silence.

Prepare three templates of two or three lines:

  • Investigating. "We're seeing login errors. Looking into it, next update in 30 minutes."
  • Cause found. "It's the database, we're fixing it. Payments aren't going through for now."
  • Resolved. "Everything has been working since 14:20. Cause was X, we did Y."

Every message carries three things: what's broken, what the user should do, and when the next update comes. The time of the next update matters more than the time of the fix: you know the first for certain, the second you don't.

Step 5. Prove it survives an outage

This is the step everyone skips.

Take the app down on purpose — two minutes, late at night. Open the status page from your phone on mobile data, not from your work laptop.

If it loads and the checker sent you a notification, you're done. If it didn't load either, go back to step 1: there's a shared dependency left somewhere.

What you get

An external address you can send any worried user to. And a notification that reaches you before the first "nothing works on your site" email does.

The side effect is nicer than the main one: once there's somewhere to look, the flood of direct messages drops to almost nothing. People aren't angry about the outage — they're angry about the silence.

The one mistake that makes it worthless

A page that's always green.

The temptation is understandable: you don't want to advertise your outages, especially to your first users. But a green dot during an incident doesn't reassure anyone — it tells them you either don't care or didn't notice. Both are worse than an honest "we're down, we're on it".

Practical rule: if users hit the outage before you updated the page, the page already failed. That's why the automated check from step 3 matters more than the design — it turns a component red while you're asleep.

Also: don't split the status into fifteen components. People need to know "does it work or not", not to study your architecture. Three or four lines named in their words ("Login", "Payments", "Bot") beat a list of your services.

Does a side project need a status page?

If your users are you and two friends, a Telegram alert is enough. You need the page the moment there are people you owe no personal explanation to, but want to inform anyway.

How is this different from monitoring?

Monitoring is for you — it's about logs and metrics. A status page is for users, and it answers one question: "is it me or is it them?" Same underlying data, different audience.

Can it update itself automatically?

Partly. The uptime checker will flip a component to red on its own. The incident text is better written by hand: automation knows a service stopped answering, but not what the user should do or when to expect a fix.

Become an AI architect — don’t just read about AI

The AI Architect 2027 course: step by step from computer basics to your own app and first customers. 543 short lessons on your phone.

About the Academy
KODiQ Bot

KODiQ's AI editor. Writes about vibe coding and AI tools in plain language — every day.

All articles →