Is a Website Down for Everyone or Just Me? Complete Diagnosis Guide

What “Down for Everyone” Actually Means (Technically)

A website is down for everyone when independent network probes across multiple Autonomous Systems (ASNs) all fail to receive a valid HTTP 200 response from the origin server or its CDN edge layer. An ASN is a collection of IP routing prefixes operated by a single administrative entity — Google LLC operates AS15169, Cloudflare operates AS13335, and Quad9 DNS operates AS19281. These networks peer via BGP (Border Gateway Protocol) but maintain entirely separate routing tables, infrastructure, and physical paths through the internet.

This definition matters because it rules out false positives. No single network failure can simultaneously break connectivity across unrelated Autonomous Systems. An ISP route flap affects only traffic transiting that ISP. A DNS cache poisoning event affects only users of that resolver. A BGP misconfiguration at a single AS withdraws only that AS’s prefixes. When all three of DownCheck’s probes — dispatched from separate ASNs — fail to reach a target within 10 seconds, the probability of a coincidental multi-AS failure is negligible: the origin server, its DNS infrastructure, or its CDN layer is genuinely unreachable.

Confirming a global outage requires querying 4 or more independent DoH (DNS over HTTPS) resolvers simultaneously — Cloudflare (1.1.1.1), Google (8.8.8.8), Quad9 (9.9.9.9), and OpenDNS (208.67.222.222) — and cross-referencing HTTP response codes with DownCheck’s multi-probe telemetry. If DNS resolution fails across all four resolvers and the HTTP probe returns a connection timeout rather than a 5xx error, the most likely root cause is a DNS infrastructure failure at the authoritative nameserver level, not an origin server crash.

What “Down Just For Me” Means and Why It Happens

A website that appears down only on your connection is experiencing a local failure — a fault in the network path, name resolution, or configuration between your device and the internet, rather than at the origin server. DownCheck’s probes reach the website successfully while your browser cannot because you are traveling a different network route. Six distinct failure classes account for the overwhelming majority of “just me” reports:

Each failure class produces a different browser error. ERR_NAME_NOT_RESOLVED points to DNS. ERR_CONNECTION_REFUSED means TCP was rejected — often a hosts file redirect to localhost. ERR_CONNECTION_TIMED_OUT suggests an ISP routing or firewall issue where traffic is dropped silently without a TCP RST packet. Matching your error code to the cause reduces the diagnosis from 7 steps to 2 or 3.

The 7-Step Diagnosis Checklist

Follow these steps in order. Each step either confirms the site is globally down (and you can stop) or isolates the local failure to a specific layer of your network stack. Completing all 7 steps takes under 5 minutes and identifies the root cause in over 95% of cases.

How DownCheck Detects Global vs Local Outages

DownCheck’s architecture is built around a 3-probe, multi-Autonomous-System design that eliminates false positives from single-point-of-measurement tools. Unlike crowd-sourced outage detectors that rely on user reports (which lag real incidents by 15–30 minutes), DownCheck dispatches active network requests the moment you submit a URL and returns a confidence-scored verdict within seconds.

The three probe results are combined into a weighted confidence score that drives the final verdict. The HTTP probe carries the highest weight because it tests actual application-layer reachability. The Cloudflare Edge probe distinguishes CDN-layer failures from origin failures. The DNS probe confirms whether the domain can even be resolved before HTTP connectivity is attempted.

When the DNS probe fails with NXDOMAIN across all 4 resolvers but the HTTP probe times out rather than receiving a connection refused, the most likely root cause is an authoritative DNS infrastructure failure — the nameservers for the domain are unreachable, preventing any client from resolving the domain to an IP. This is distinct from an origin server crash, where DNS resolves correctly but the TCP connection to the resolved IP fails. DownCheck surfaces this distinction explicitly in the probe detail panel so you can identify whether to contact the site’s DNS provider or hosting provider.