Your Website Is Down. The First 30 Minutes — a Small-Business Playbook
Down for everyone or just you? Host, DNS, SSL, or last night's plugin update? A calm, ordered checklist for the worst half hour — and how to hear about the next outage before a customer tells you.
Your Website Is Down. The First 30 Minutes — a Small-Business Playbook
Someone texts you: "your site's not loading." Your stomach drops, you type the URL, and now you're staring at an error page too. Every minute is a customer bouncing to a competitor — and you're not sure whether to blame the host, the domain, the SSL thing, or that plugin you updated last night.
This is the calm version of the next 30 minutes: what to check, in what order, and how to be the first to know next time instead of the last.
Minutes 0–5: confirm it's actually down
Don't debug what isn't broken.
- Check from a second network. Phone on cellular data (Wi-Fi off). Ask a friend in another city, or use a "down for everyone or just me" checker. If it loads for them, the problem is your network or DNS cache — not the site.
- Try incognito. Rules out a poisoned browser cache.
- Try both
https://and the bare domain. Different failures answer differently.
If it's down for everyone: keep going.
Minutes 5–10: read the error like a clue
The message on screen narrows it to one suspect:
| What you see | What it usually means |
|---|---|
DNS_PROBE_FINISHED_NXDOMAIN / "server not found" | DNS or the domain itself — check the registration hasn't expired |
| "Your connection is not private" / SSL error | The SSL certificate expired or didn't renew |
| Error 500 / "critical error on this website" | PHP or a plugin crashed — often last night's update |
| Error 502 / 503 / 504 | The host — server overloaded, crashed, or in maintenance |
| Blank white page | A fatal PHP error with error display off — same family as the 500 |
| Loads forever, then times out | Server up but drowning — traffic spike, runaway process, or attack |
Minutes 10–20: the usual suspects, in order
- Your host's status page. If they're having an incident, your debugging is done — bookmark their page and wait it out (fastest resolution of all, and completely out of your hands).
- Domain and DNS. Look up the domain's expiry. An expired domain takes hours-to-days to fully recover — this is the one to catch early.
- SSL certificate. If the cert expired, your host can usually re-issue in minutes (most auto-renew — when the renewal breaks, nobody notices until it lapses).
- The last thing you changed. Plugin update? Theme edit? WordPress will often email the admin a recovery mode link when a plugin fatals — that link lets you log in and deactivate the broken plugin even while the front of the site is down. No email? Renaming the plugin's folder over SFTP does the same job.
Minutes 20–30: call the right people
- Host issue (5xx, status-page incident): open a ticket / hop on host chat. Say the words "my site returns a 502" — precise beats angry.
- Code issue (500 after an update): your developer, with the plugin name and the time it started. If revenue is on the line, roll back first and diagnose later — hosts with daily backups can restore in minutes.
- Domain/DNS issue: your registrar, now. This one gets slower the longer it sits.
After it's back, write yourself two sentences: what broke, what fixed it. The next outage will rhyme with this one.
The part that decides how bad this gets: who found out first?
Everything above assumes the worst version of the story — a customer told you. The gap between "site went down" and "you knew" is pure loss: bounced orders, missed leads, and the special embarrassment of hearing it from someone else.
Closing that gap is exactly what FlowAlert is for, from both directions:
- Uptime checks visit your site from the outside every few minutes. Three consecutive failures and
site.downpages whoever you've put on call — acknowledged, or escalated to a backup until someone owns it. The recovery pings you too. - The WordPress plugin watches from the inside: it pushes an alert the moment WordPress hits a fatal error and enters recovery mode, and its every-10-minutes check-in acts as a dead-man's switch for the failures that don't answer at all.
With both directions on, "the first 30 minutes" starts when the site goes down — not when a customer finally tells you. That half hour is the difference between an incident and an apology.
