How to Monitor Your Internet Connection (and Prove It Drops)

The maddening thing about a connection that drops

You are on a video call and it freezes. By the time you have said "can you hear me?" three times, it's back. Or a big upload fails at 90% for the second time, and you have no idea why. Or your ISP insists the line is "perfect" while you know — you know — it goes down every afternoon around three. The frustrating part isn't just the dropout; it's that it never happens while anyone official is looking, and you have nothing to point to except a feeling.

The fix is to stop relying on feelings and start collecting facts. A connection monitor watches your internet continuously, records the exact moment it drops and the exact moment it returns, and — if you want — makes a noise so you notice in real time instead of ten minutes later. This guide explains how monitoring actually works (including the surprisingly common way your own computer misleads you about being "online"), how to set up an audible alarm, and how to build the downtime log that finally makes your ISP take you seriously. You can follow along with our free Internet Connection Monitor, which does all of this in your browser with nothing installed and nothing uploaded.

Why "online" doesn't mean online

Here is the single most important thing to understand, because it explains half of all "my internet is weird" experiences. The connectivity indicator on your device — the Wi-Fi icon, the browser's online status, the little "connected" label — is not checking whether you can reach the internet. It is checking whether your device has a link to your router. Those are completely different things.

Picture the chain: your laptop connects to your router, and your router connects, through your ISP, to the wider internet. The indicator only watches the first hop. So when the failure is further out — your ISP's uplink drops, the fibre in the street has a fault, the modem loses sync — your laptop is still perfectly "connected" to a router that itself can't reach anything. The icon stays green. Pages stop loading. And you waste twenty minutes restarting a laptop whose Wi-Fi was never the problem.

Programmers hit the same trap: browsers expose a navigator.onLine flag, and countless apps trust it, only to discover it reports "online" while every actual request fails. The lesson is universal — the only way to know you can reach the internet is to actually try to reach it. That is exactly what an active monitor does, and it is why active monitoring catches outages the status icon sleeps right through.

How active monitoring works

Active monitoring is refreshingly simple in principle. On a short, regular interval — every few seconds — the monitor makes a tiny network request to a host that is essentially always up, and times how long it takes to respond. Three outcomes follow from each check:

• →Fast response: you're online, and the round-trip time is your current latency.
• →Slow response: you're online but the connection is degraded — the number climbing is your early warning.
• →No response / timeout: the request failed, and after a check or two the monitor declares you offline and starts the clock on an outage.

The interval is a trade-off worth understanding. Checking every 3 seconds catches a dropout almost the instant it happens but makes more requests; every 30 seconds is gentler but might miss a brief blip entirely. For watching a call or an upload, a short interval is right; for background peace-of-mind over hours, a longer one is kinder. Our monitor lets you pick 3, 5, 10 or 30 seconds so you can match the check rate to what you're actually doing.

This is the same core idea behind professional uptime services, just pointed at your own connection instead of a website. If your interest is the other direction — checking whether a site is up rather than your link — our website uptime checker and its guide cover that case, and our speed test measures raw throughput when you want megabits rather than milliseconds.

Reading latency and jitter

Latency — the round-trip time for a request, measured in milliseconds — is the number the monitor graphs, and learning to read it tells you a lot. As a rough guide: under 50 ms is excellent, 50–100 ms is comfortable for browsing and calls, and consistently above 150–200 ms starts to feel sluggish, with video calls stuttering and games becoming unplayable. But the steady number is only half the story.

The other half is jitter — how much the latency jumps around between checks. A connection that holds a steady 80 ms feels smoother than one that bounces between 20 ms and 200 ms even though the second might have a lower average, because real-time applications hate unpredictability. This is why the monitor's latency graph matters more than any single reading: a wall of even bars is a healthy connection, while a jagged, spiky graph reveals instability that a one-off speed test would completely miss. When you're diagnosing "the call keeps glitching but the speed test says I'm fine," the spiky graph is your answer, and the red bars where checks failed entirely are the dropouts themselves.

Setting up a downtime alarm

Detecting a drop is only useful if you notice it, and staring at a graph all day is not a plan. That's what the alarm is for: the moment monitoring flags you offline, it plays a sound until you're back online or you silence it. Suddenly a dropout during an important upload, a live stream, or a work-from-home day announces itself instead of costing you ten silent minutes.

Our monitor generates a handful of built-in tones — a sharp beep, a rising-and-falling siren, a gentle chime, a steady pulse — synthesized on the fly, so there's no audio file to load. You can also upload your own sound: a voice memo saying "internet's down," a song clip, anything that will get your attention, and it stays on your device. Crucially, there's a Test button next to the volume slider, because an alarm you've never heard is an alarm you can't trust. Press it, confirm it's loud enough and it's the sound you want, and only then rely on it. One browser quirk to know: audio is blocked until you interact with the page (to stop websites blaring sound on load), so pressing Start monitoring or Test is what unlocks the alarm for later automatic playback.

For a quieter alternative, the monitor can also fire a desktop notification on both drop and recovery, so a glance at the corner of your screen tells you the state without any sound at all — handy in a shared room or an open office.

Building the evidence: the downtime log

This is where a monitor earns its keep against an ISP. Every time your connection drops, the monitor records the timestamp; when it recovers, it records the duration. Over a few days you accumulate something no support agent can dismiss: a dated list of outages with exact times and lengths, a running total of downtime, and a count of how many times it happened. "My internet feels unreliable" is easy to brush off; "the connection dropped fourteen times between Monday and Wednesday, for a total of 47 minutes, mostly between 2 and 4 pm" is a pattern your provider has to engage with.

The log persists locally in your browser, so it survives closing the tab and comes back when you return — you're building a record over time, not starting fresh each session. A few tips make the evidence stronger. Run the monitor during the window when problems actually occur, on a device wired to the router if you can, to rule out Wi-Fi as the variable. Note whether outages cluster at particular times, which points at congestion or a scheduled process rather than a random fault. And pair the log with our speed test results from the same period: "drops fourteen times a day and delivers half the speed I pay for" is a far harder complaint to wave away than either fact alone. If a specific service rather than your whole connection seems to be the problem, the "is it down or just me?" approach helps you tell the two apart before you call anyone.

What a browser monitor can — and can't — do

It's worth being clear-eyed about the boundaries, because they explain when a web tool is the right choice and when it isn't. A browser-based monitor is perfect for watching your connection while you work: it needs no installation, runs on any device, keeps everything private on your machine, and is ready in one click. Its honest limitations both come from the browser sandbox. First, it can only run while its tab is open and, for second-by-second accuracy, foregrounded — browsers deliberately throttle background tabs to save battery, and stop them entirely when closed. So a web monitor is a "watch while I'm working" tool, not a 24/7 unattended sentinel.

Second — and this is the question people always ask — a browser tool cannot show you which applications are using your network, list open ports, or capture and inspect packets the way Wireshark does. That isn't a missing feature; it's a security wall. Browsers are sandboxed precisely so that a web page can never see other programs' traffic or your system's network internals — if any website could do that, it would be a catastrophic privacy hole. Packet capture, per-app network monitoring and port inspection require a native desktop application running with system privileges (and, for packet capture, a special driver and administrator rights). Those are legitimate tools, just fundamentally not something any web page can be. Knowing this boundary saves you hunting for a browser tool that, by the laws of the platform, cannot exist. For everything on the right side of that line — is my connection up, how's my latency, when did it drop, alert me when it does — the browser monitor is exactly the tool.

A practical monitoring routine

For a flaky connection you want to fix: run the monitor at a 5-second interval on a wired device across the hours when drops happen, over two or three days. Let the log build, note the timing patterns, run a speed test at both good and bad moments, then take the combined evidence to your ISP.

For an important call or upload right now: start monitoring at a 3-second interval with the alarm on and volume tested, so the instant anything drops you know immediately and can pause, reconnect, or switch to a backup connection like a phone hotspot.

For background peace of mind: a 30-second interval with desktop notifications instead of sound keeps an unobtrusive eye on things while you work, flagging only genuine drops.

In every case the principle is the same one this guide opened with: replace the vague feeling that "the internet is being weird" with a measured, timestamped, provable record. Once you can see exactly when and how often your connection fails, you can fix it, route around it, or make someone else fix it — and you can finally trust the green icon, or catch it in the lie.

Frequently Asked Questions