What an Internet speed test measures—and what it cannot prove

A speed test typically exchanges data between your device and a test server, then estimates how that exchange performed over a period of time. The result describes that particular connection, at that particular moment, under the conditions then in place: the device, the local network, the provider’s access network, and the participating server. It is not a direct reading of one fixed, unchanging property called “Internet speed.” The International Telecommunication Union treats performance measurements as methods that depend, among other things, on how the test is conducted. It is therefore useful to read a figure with its context in mind. ITU-T Q Suppl. 71 addresses measurement methods in fixed and mobile networks.

The route may include the device itself, its connection to the router, the home link, and several network segments on the way to the selected server. If any of these elements limits the exchange, the test may record a value below the maximum advertised for your access service. A particular server or route may also fail to represent performance when connecting to other services. A result is therefore a useful clue for guiding a diagnosis, but it does not, on its own, identify which component explains slow performance or conclusively establish how the connection will perform in every circumstance.

First decide what question you are trying to answer. If you want to find out approximately how fast your home access service can be, a direct Ethernet connection to the router reduces the influence of the wireless signal. If the problem occurs when using a laptop in a particular room, testing over Wi‑Fi in that room may be more relevant: it shows the performance experienced there, even though it does not isolate the cause. These are different questions and should not be mixed when comparing results. DIGI’s speed-test help page recommends connecting to the router via Ethernet, temporarily disconnecting other devices, and closing applications to obtain more accurate results. This is the provider’s recommendation, not a guarantee that the test will reproduce every real-world usage condition.

Make a simple note of the device used, whether it was connected by cable or Wi‑Fi, and where it was located. Close large transfers, cloud backups, and downloads that are not part of the situation you want to measure. If you suspect the issue occurs when the household is busy, also test under those conditions and record them. There is no need to eliminate all network activity permanently: the aim is to distinguish a controlled test from one that reflects everyday use. If you change the device, location, and connection type at the same time, a difference between the figures will not tell you which change made a difference.

Download is the rate at which data arrives from the test server at your device during the measurement. It matters for tasks such as receiving files or streaming content, but it does not necessarily equal the rate you will get from every service: the route, remote server, and device load can also have an effect. The figure is normally shown in megabits per second (Mb/s or Mbps). Do not automatically confuse that with megabytes per second, which is a different unit: one byte contains eight bits, before other transfer details are taken into account. A high download result does not prove that every other aspect of the connection is equally good.

Upload estimates the rate at which your device sends data to the test server. It can matter when sending files, backing up content, or streaming video, but the figure still describes that particular exchange. Downloading and uploading are different operations; do not assume they will produce the same value. Some plans specify symmetrical speeds and others do not, so compare the measurement with the terms of your particular plan rather than a general expectation. If upload is much lower than download, repeat the test under stable conditions and check whether other applications are sending data before concluding that there is a fault.

Latency refers to how long a communication takes to travel along the test route and return, and is usually expressed in milliseconds. It does not indicate how much data is transferred per second. A connection may have a high download rate and, at the same time, latency that is unsuitable for an activity sensitive to delay; it may also have low latency and limited throughput. The names and level of detail of the metrics depend on the tool. FAST.com, for example, says its test calculates download speed and that its expanded information view shows upload and latency, with measurements associated with download and upload.

The latency in a result depends on the route to the chosen server and the conditions during the measurement; it cannot simply be applied to every Internet destination. When a tool shows latency during download or upload, those figures let you observe responsiveness while traffic is moving in that direction, as well as the measurement without that load. This distinction helps explain why a connection may feel less responsive while data is being transferred, but it does not turn the test into a complete diagnosis of every application. For comparisons, use the same tool and similar conditions, and look at variation across repeated tests rather than reading too much into one result.

Compare the figure with the speed stated in your contract and the conditions that apply: access technology, the type of speed specified, and the measurement method. A Wi‑Fi test in a distant room does not assess the provider’s network segment in isolation, just as a wired test near the router does not necessarily describe the experience on every device in the home. The difference may lie before or after the router. It is best to separate these possibilities with comparable tests first, rather than immediately choosing an explanation.

The CNMC offers an Internet speed test that its own page describes as official and free. It can serve as an alternative check, but the information on the cited page does not allow us to attribute a detailed technical procedure to it here or to claim that its results alone constitute formal proof of a breach. The CNMC service does not replace reading your contract or checking its terms. If results remain low across several controlled measurements, keep the date, time, tool, connection type, and figures, then contact your provider and request a check. The available evidence establishes no universal threshold for declaring a connection faulty on the basis of a single test.

Take several measurements at separate times, preferably with the same device and tool. Repeat one set over cable and another over Wi‑Fi, keeping the location and household activity reasonably consistent when comparing each set. If the wired result is stable but Wi‑Fi performance drops in a particular area, the wireless network or device are aspects worth checking; that difference points to possibilities but does not, by itself, prove the cause. If performance also drops over cable, the test still cannot distinguish between the equipment, router, provider, server, or measurement route.

Test at different times if the problem seems intermittent, and record whether other devices were using the connection. Do not treat the highest figure in a set as typical performance or the lowest as automatic proof of a permanent fault. A repeated, documented measurement is more useful for diagnosis than an isolated screenshot because it can reveal patterns and show the conditions in which they appear. Even so, it remains an observation of traffic between a client and a server, not a complete network audit. The cautious conclusion is limited: the test shows how that route performed at that moment; orderly comparisons help decide what to check next.