How to show your ISP it's not your end
"Have you tried restarting your router?" Support lines are built to rule out the easy causes first, and they're right to: often it is something at home. If it isn't, the way through is evidence. Here's how to gather it.
1. Rule out your own network first
Do these before you call, and say that you've done them. It saves a round of back and forth, and it's the first thing you'll be asked.
- Test with a cable. Plug a computer straight into the router. If the problem goes away, it's the Wi-Fi, not the internet connection.
- Restart the router once. Then note whether the problem comes back, and when.
- Check it isn't one device. If only one computer suffers, look at that computer first.
- Check nothing on your network is using it all. A large upload, such as a cloud backup, can fill the connection and cause exactly the symptoms you're chasing.
2. Watch the right things, at the same time
Measuring one website tells you something is wrong, not where. Watch several points along the way, all at once, so you can see which ones suffer together:
| Watch | Why |
|---|---|
| Your own router | If this shows loss, the problem is inside your home. |
| Your provider's first router (usually hop 2 in a traceroute) | The start of their network. |
| One or two large public services (for example 1.1.1.1 and 8.8.8.8) | Well-connected and reliable: if they suffer too, it isn't one website. |
| The thing that's actually giving you trouble | So the evidence matches your complaint. |
Then read the pattern:
- Your router clean, everything from the provider onwards bad: the problem is on the line or in your provider's network. This is the pattern to take to them.
- Your router bad too: it's inside your home. Look at the router, cabling or Wi-Fi.
- Only one destination bad: it's probably that service, or the route to it, not your connection as a whole.
Add a traceroute to the troubled destination, and look for the hop where loss starts and carries on to the end. Which hop is causing packet loss? explains how, including why one router showing loss on its own doesn't count.
3. Collect it over time
Intermittent problems are the hardest to get fixed, because they're rarely happening while someone's looking. Leave the monitoring running for days if you can, so it catches the bad spells as they happen. Two things make the evidence much stronger:
- Times and dates of every drop, so it can be matched against their own records.
- A pattern, such as "every evening from about 7pm", which often points to a link that fills up at busy times.
4. What to send them
Keep it short and factual. Numbers, times and what you've already ruled out get escalated; "it's been rubbish all week" doesn't. Something like:
Since 14 March I've had packet loss of 15–30% most evenings between roughly 7pm and 11pm, measured continuously. My own router shows no loss at the same times, so it isn't my home network. Loss starts at your router (hop 2) and continues to every destination I monitor, including 1.1.1.1 and 8.8.8.8. I've tested with a wired connection and restarted the router. Graphs and the full log are attached. Please can this be investigated as a line or network fault?
Attach graphs of the bad periods and the traceroute showing where the loss starts. If you're asked for more, a packet capture is the strongest evidence of all: it shows your computer sending each request and no reply coming back.
BadHop is built for exactly this. Watch your router, your provider and any number of destinations side by side, for as long as you like; every alert and recovery is logged with its time; graphs can be saved as images for your ticket; and packet capture exports a standard .pcap file. See the features, or download it free.