To test business internet speed properly, connect a computer by cable as close to the carrier's equipment as you can, stop other heavy traffic on that machine, and run several tests in both directions at different times of day. A single test from a laptop on Wi-Fi measures the Wi-Fi, the cabling and the laptop as much as the circuit. If wired results at the edge of your network match what you bought, the circuit is doing its job and the slowdown is somewhere inside the office.
A lot of "the internet is slow" calls start with a screenshot of a speed test taken on someone's phone. Sometimes that screenshot is right. More often it is measuring something other than the line you pay the carrier for, and a ticket opened on the strength of it ends with a technician reporting that the circuit tests clean. A few minutes of method saves that round trip, and it tells you whether to spend money on a bigger circuit or on the network inside your walls.
Why a quick test so often misleads
A speed test measures the slowest link between your device and a test server somewhere on the internet. Plenty of things sit on that path besides your circuit:
- Wi-Fi. Distance, walls, interference and the age of both the access point and the laptop cap what you see. It is common for a Wi-Fi test to show a fraction of a fiber circuit's speed while the circuit itself is fine.
- Everyone else in the office. If a backup is running or half the team is on video calls, your test only gets what is left over.
- The device. An older network adapter, a VPN client, or security software inspecting every packet can hold a laptop well below the line rate.
- Your firewall. Firewalls are rated for throughput, and features such as intrusion prevention and TLS inspection reduce it. A firewall bought for a 200 Mbps line quietly becomes the ceiling on a 1 Gbps circuit.
- The test server. A busy or distant server, or one reached over a congested route, can report a low number that has nothing to do with you.
None of this makes speed tests useless. It means one test, taken anywhere, proves very little on its own.
Before you test: a checklist
- Use a laptop or desktop with a gigabit (or faster) wired network port, and check the adapter's link speed in the operating system's network settings.
- Use a known-good patch cable. A damaged or very old cable can drop the link to 100 Mbps without any warning on screen.
- Disconnect from any VPN, and pause cloud sync, updates and large downloads on the test machine.
- Take the first baseline at a quiet time, such as early morning, then repeat at your busiest hour.
- Know what you bought: the download and upload speeds on the order, and whether the product is dedicated or shared.
- Write down the time, where you plugged in, the server used and each result. A screenshot with no time or place is weak evidence.
Where you plug in decides what you measure
Testing at several points, moving closer to the carrier's equipment each time, is the quickest way to find a bottleneck. Each step removes one suspect from the list.
| Test point | What the path includes | What it tells you |
|---|---|---|
| Phone or laptop on Wi-Fi | Wi-Fi, office switches, firewall, circuit | What staff experience, but not where a problem sits |
| Wired at a desk | Desk cabling, switches, firewall, circuit | Whether Wi-Fi is the problem |
| Wired at the main switch | Firewall and circuit | Whether floor cabling or edge switches are the problem |
| Wired to a spare firewall LAN port | Firewall and circuit | The best result most offices can measure without downtime |
| Directly at the carrier handoff | The circuit alone | The circuit's own performance; usually done at install or with the carrier, because it means unplugging your firewall |
Do not unplug the firewall during business hours to test at the handoff. Testing directly on the circuit also exposes the test machine to the internet with no protection in front of it. If you need that measurement, arrange it with the carrier or your IT provider, out of hours.
Running the test
Choose the server deliberately
Browser-based tests usually pick a nearby server automatically. Run the test against two or three different servers, including one hosted on your own carrier's network if the tool lets you choose. If one server is consistently low and the others are fine, the issue is that server or the route to it, not your line.
Test both directions
Upload matters for video calls, cloud backup and sending files. On a dedicated fiber circuit the upload result should be close to the download result, because the service is symmetrical. On coax, upload is normally a small fraction of download by design, so compare it with the plan's upload figure rather than the download. Our guide to symmetrical vs asymmetrical internet explains why that gap exists.
Repeat across the day
Run a set of tests early, mid-morning, after lunch and late afternoon, over two or three days. A dedicated circuit should look much the same at every hour. A shared service may dip at busy times, which is normal within limits. A result that falls only when your own office is busy points to your own usage filling the line.
One connection or many
Most speed tests open several connections at once, which fills a fast line easily. A single file download uses one connection, and over long distances one connection often cannot reach the full line rate because of how TCP waits for acknowledgements before sending more. If a multi-connection test shows full speed but one download from a distant server is slower, the line is probably fine. Technical teams use tools such as iperf against a server they control, so they can set the number of streams; ask whether your carrier has a test server you can use.
What a good result looks like
No circuit tests at exactly its line rate. Ethernet, IP and TCP each add headers to every packet, so the data a test can measure is a little below the circuit speed. A gigabit circuit commonly tests in the 900s rather than at 1,000, and a 500 Mbps circuit somewhat below 500. That gap is overhead, not a fault.
The bigger question is consistency. On dedicated internet access, your committed bandwidth should be there at every hour, in both directions. On shared fiber or coax, the plan speed is an "up to" figure, and some variation at busy times is part of the product. A result that sits far below the plan for long stretches is worth raising on either kind of service.
Reading the results
| What you see | Likely cause | Next step |
|---|---|---|
| Wired results stop at about 94 Mbps | A link negotiated at 100 Mbps: bad cable, old switch port or an adapter setting | Swap the cable and the port; check link speed on the switch |
| Wired is fine, Wi-Fi is slow | Coverage, channel congestion or old access points | Look at the Wi-Fi before touching the circuit |
| The same ceiling at every wired point, below the plan | Firewall throughput limit, or a circuit problem | Check firewall load during a test, then test at the handoff with the carrier |
| Fine early morning, slow mid-afternoon | Your own office filling the line, or congestion on a shared network | Check the firewall's usage graph at the slow times |
| Upload far below download on dedicated fiber | Configuration, firewall policy or a provisioning error | Raise it with the carrier; DIA should be symmetrical |
| Fast tests, but one cloud app is slow | That service, or the route to it | Test against that service; check its status page |
A worked example
Hypothetical numbers for illustration, not a real client or a real test.
A 30-person office buys a 500 Mbps dedicated circuit, and staff complain that it feels slow. The office manager's phone test on Wi-Fi shows 160 Mbps down. A laptop wired at a desk shows 94 Mbps in both directions, every time. The same laptop wired into a spare port on the firewall shows 470 Mbps down and 465 Mbps up at 8am, and about 440 Mbps each way at 3pm while the office is busy.
That gives three findings. The circuit is delivering what was ordered; the gap below 500 is protocol overhead plus normal office traffic. The 94 Mbps ceiling is a link stuck at 100 Mbps somewhere between that desk and the main switch, most likely an old cable or a damaged patch panel port. And the Wi-Fi result points to access points that need attention. A bigger circuit would have fixed none of it, and buying one would have added monthly cost while leaving every complaint in place.
When to call the carrier, and what to send
If wired tests at the firewall are well below the plan across several days, and the firewall is nowhere near its limit, open a ticket. Send:
- The circuit ID or account number.
- A short table of results with date, time, test point, server and both directions.
- Confirmation that the tests were wired, with nothing else heavy running on that machine.
- Whether the shortfall is constant or appears at certain hours.
That evidence usually moves a ticket past the first-line script. The carrier may run its own test from its network to your equipment, or ask for a window to test at the handoff. If you buy through an agent, they can open and chase the ticket for you, which is a large part of the value of having one person who knows the account.
Speed is one number among several
A line can pass every speed test and still make calls choppy, because voice and video care about delay and consistency more than raw throughput. If the complaint is about calls freezing rather than downloads crawling, look at latency, jitter and packet loss instead; our guide to latency, jitter and packet loss covers how to measure them. If your tests show the line running full at busy hours, the question becomes sizing, which do you need gigabit internet works through. And for offices that would rather have this watched continuously than tested by hand, a managed network service tracks usage and performance around the clock.
Questions to ask your provider
- What download and upload speeds is this circuit provisioned for, exactly?
- Is the bandwidth dedicated, or shared with other customers at any point?
- Do you have a test server on your network that I can use for wired tests?
- Can you run a test from your side to my equipment and share the results?
- What does the service description or SLA say about throughput, and how is it measured?
- If my tests show a shortfall, what information do you need to open a ticket?
Get a second opinion on the circuit
If your tests point to a circuit that is too small, too slow or simply overpriced, we can price the alternatives at your exact address. Several carriers bid, we bring back the best offer, and the carrier pays us, so there is no markup on your bill. If your current deal turns out to be the best one, we will say so. The quote is free, carries no obligation and usually comes back the same day. Send us your address, call 478-758-8091 or text (347) 870-0965.
