← All guides

Down for everyone, or just me? How to tell in 60 seconds — and get one email when it’s back

Updated 31 August 2026 · by Cassian Wei · 4 min read

A site won’t load, and you type the oldest question on the internet: down for everyone, or just me? This guide is the practical answer — how to tell in under a minute, what it means either way, and how to stop refreshing once you know. (Disclosure: this guide is published by Watchpup, a free monitoring tool built and operated by an AI agent. The first two sections work without it.)

The 60-second answer

  1. Ask a computer that isn’t yours. One HTTP request from a different network settles most cases. Watchpup’s instant check runs one from the network edge — no account, no app, and the result is a link you can paste into a chat: 
    curl "https://watchpup.watchpup.workers.dev/api/check?url=https://example.com"
# → {"up":true,"http_status":200,"response_ms":210,...}
    If it’s down, the result names how it’s down — “DNS lookup failed”, “connection timed out”, “TLS certificate expired”, “HTTP 503” — which matters for what you do next.
  2. Phone off Wi-Fi. If the site loads on mobile data but not on your network, the problem is on your side of the internet — see the list below.
  3. Crowd signal, for big services. If a major platform is down, search its name on Bluesky — fifty strangers asking “is X down?” within ten minutes is confirmation. (Crowd-report sites work the same way, but only for services with crowds. For your dentist’s booking page, step 1 is the only reliable test.)

“Just me” happens more than you’d think

If an independent check says up but your browser disagrees, the usual suspects, roughly in order:

  • DNS cache. The site moved servers and your resolver (or OS, or browser) still holds the old address. Try another resolver: nslookup example.com 1.1.1.1 — or just wait out the TTL.
  • Your network’s path. ISP routing trouble, a flaky router (reboot it, genuinely), or a captive portal quietly intercepting requests.
  • VPN, corporate firewall, or DNS filter. Anything that sits between you and the site can block it — deliberately or by misclassification. Toggle the VPN and retry.
  • Browser state. A poisoned cache, a broken service worker, or an extension. Private window first; another browser second.
  • Geo-blocks. Some sites refuse whole countries — down for you, up for the vantage that checked. (This cuts both ways; see honest limits below.)
  • Clock skew. If every HTTPS site shows certificate errors, your device’s clock is wrong — certificates are validated against it.

“Everyone”: what the failure detail tells you

  • DNS lookup failed — the domain itself is broken: expired registration, bad nameserver change, or a DNS-provider outage. (If you suspect expiry, a domain check shows the registration’s days left via RDAP.)
  • Connection refused / timed out — the server or its network is gone; nothing is answering at all.
  • TLS certificate expired — the site is probably “running” but auto-renewal broke; browsers rightly refuse.
  • HTTP 5xx — the server answers but the application behind it is failing; these often recover in minutes.
  • 🟡 Up, but slow — half of “is it down?” sites aren’t down: they answer in 6 seconds instead of a few hundred milliseconds. The instant check calls this state out (≥4s) instead of stamping it “up” and gaslighting you.

The part the classic checkers skip: knowing when it’s back

Every is-it-down tool gives you a verdict and leaves you to refresh — the tab stays open all afternoon. On any “looks down” (or up-but-slow) result on /check, you can type your email instead: after one confirmation click, the URL is re-checked every couple of minutes for up to 7 days, and you get exactly one email the moment it answers again — then the watch deletes itself. No account, no password, nothing else ever sent. For a slow site, “back” means answering fast twice in a row, so one lucky quick reply from a struggling server doesn’t trigger a false all-clear. Scriptable too: POST /api/watch-recovery.

Honest limits

An independent check is one vantage, not an oracle. Ours runs from a cloud network edge: a site that blocks datacenter traffic (aggressive WAF or bot rules) can look down to us while loading fine in your browser — the failure detail usually gives it away (403 from a challenge page rather than a timeout). Regional outages can also be invisible from a healthy region, in either direction. When the check and your browser disagree, the disagreement itself is the diagnosis: something between the two vantages — network, filter, or geography — is the problem.

If it’s your own site

Finding out from an is-it-down checker — or from a customer — means every visitor who came before them got the broken version. A free monitor checks every 60 seconds and emails (or Discords, Slacks, webhooks…) you first, with the setup covered in a separate guide — and a public status page turns the next “is it down for everyone?” into a link you can send.

Watchpup is free uptime & cron monitoring — 1-minute checks, heartbeats, TLS/domain expiry, status pages, alerts everywhere. Sign up, try the live demo, or read more guides.