Before any cable is touched, one question decides where the fault lives: how many machines are affected? Scope first, tools second.
Network faults punish people who start fixing before they start counting. If one computer has no internet while its neighbours browse happily, the router and the internet line are innocent — the fault is inside that PC, its cable, or the switch port it hangs from. If every computer is down, no driver reinstall on any single PC will help, because the fault is shared: router, switch, or the line coming into the building. This guide turns that scope idea into a method — the scope count, the ping ladder that walks the fault outward step by step, and the local faults that imitate network death.
👆 Example scopes. One minute of counting replaces an hour of repairing the wrong box.
Ask three counting questions, in this order:
The answers draw a boundary around the fault. One wired PC down with Wi-Fi fine in the same room: the boundary is that PC and its wire. All wired down but Wi-Fi working: the boundary is the switch. Everything down including phones: the boundary is the router or beyond it — call the ISP after one router restart, and stop reinstalling network drivers on healthy PCs.
From the sick PC, ping outward in steps. Each reply clears a rung; the first silence names the failing stretch:
ping 127.0.0.1 — the PC's own network stack answersping the PC's own IP (from ipconfig) — its network adapter is aliveping the gateway (usually the router, like 192.168.1.1) — the cable, switch and local network path workping 8.8.8.8 — the internet itself is reachable by numberping google.com — DNS name resolution worksFails at rung 3: stay local — cable, port, adapter. Passes rung 4 but fails rung 5: the connection is fine and it is a DNS fault — set a known DNS or fix the router's. The ladder takes two minutes and never guesses.
When the scope is one PC, the imitators queue up. An IP address starting 169.254 means the PC asked the router for an address and got no answer — cable, port or the router's DHCP service, not the internet. A duplicate IP after someone set a static address by hand makes two PCs knock each other offline in turns. A software firewall or a freshly installed security suite can block a perfectly good connection — test with it paused for one minute, honestly, then re-enable. And the humble classic: the cable that clicks into the switch but not into the PC. Each of these presents as no internet; each lives entirely inside the one-PC boundary the scope count already drew.
Whole-network deaths have their own short list. A router that needs a restart every few days is ageing or overheating — note the pattern instead of just rebooting. A switch that takes one floor down has failed or lost its uplink cable — check its lights and reseat the uplink. Sudden total loss after rain or roadwork nearby is the ISP's physical line more often than any equipment of yours. The professional habit is evidence: note the time, which scopes died, and the router lights' state, then call the ISP with facts. Their first question — is it just you — is the scope count you already did.
Watch network faults scoped and ladder-tested in the video — one-PC faults, switch faults and a whole-office outage separated live, step-by-step.
▶️ Watch Full VideoInside your boundary: your PC, cable and port. Run ipconfig — a 169.254 address means the router never gave you an address, so reseat or swap the cable and try another switch port. Then walk the ping ladder to see which rung fails.
DNS — the phone book of the internet. Your connection works; name lookup does not. Set a public DNS such as 8.8.8.8 on the PC to confirm, then fix the DNS setting in the router so every device heals at once.
Partly — it is the wireless side only: the access point or the router's radio. Restart the access point, check whether the Wi-Fi name is even visible, and test a phone near the AP. The internet line and wired network are already proven innocent by the working PCs.