What do your internet speed test results mean?

Four numbers describe different parts of one test route. Use them together, then repeat under comparable conditions before deciding whether your device, Wi-Fi or plan needs attention.

By A. Ash · Updated October 30, 2026 · About 8 minutes

Mbps and ms are different

Mbps means megabits per second, a rate of data transfer. It is different from MB/s, megabytes per second: eight bits make one byte, before accounting for overhead. A 100 Mbps reading is therefore roughly 12.5 MB/s as a simple unit conversion, not a promise that every download will transfer at that rate. The remote server, protocol, device and route also matter.

Ping and jitter are measured in milliseconds. A lower ping to this site can indicate a more responsive path, while lower jitter indicates more consistent timing. There is no universal “good” threshold that predicts every video call or game: each service reaches a different destination and has its own buffering and performance needs.

Compare two tests fairly

1. Write down the plan tier. Check both advertised download and upload, the access technology and whether the number is a maximum or a guaranteed minimum under your terms.
2. Get a wired baseline. If possible, use Ethernet on a capable device, pause large transfers and VPN traffic, and run AI Speed two or three times. Record date and time.
3. Test your usual Wi-Fi spot. Use the same device without moving other variables. If Wi-Fi is much slower while wired remains stable, examine signal, router placement and the device before changing providers.
4. Repeat at the problem hour. Compare a quiet period with the time calls or streaming fail. Record other household usage. A repeated pattern is more useful to support than one unusually low reading.
5. Cross-check a second test. Different services use different servers, routes, parallel flows and data volumes. Agreement adds context; disagreement is a reason to investigate the route or endpoint, not automatic proof of throttling.

What pattern do you see?

Choose the next test from the result—not from a guess

Use the first comparison to change only one condition at a time. This does not identify the fault automatically; it creates stronger evidence for the next decision.

For support, save several comparable readings rather than the single worst result. The connection worksheet records date, time, device, Ethernet or Wi-Fi, other traffic and the four measurements without sending the record to AI Speed.

What if the number is below your plan?

A retail tier and a browser reading answer different questions. Your measured number includes the device, Wi-Fi or Ethernet link, router, other traffic, route and the test endpoint. Start with the wired baseline and several dated runs. If they remain far below the subscribed tier, ask your provider what to expect at the modem or optical terminal and provide the conditions you recorded. Read why tests disagree before relying on a single result.

The CRTC’s universal service objective calls for access to at least 50 Mbps download and 10 Mbps upload with an unlimited data option. It is an availability objective, not a guarantee that every device or test route must return 50/10. Read the CRTC explanation ↗.

What this test cannot tell you

AI Speed measures the path between your browser and this site’s hosted endpoint at one moment. It cannot identify a fault within your provider’s network, certify your access line’s maximum speed, predict a game server’s latency, or produce a representative city or ISP average from your one result. The IP location lookup started with a test, or requested manually, estimates location and ownership of an IP range; it is not a precise address.

To shop, confirm service at the complete civic address and unit number. Use the provider comparison and regional directory as research starting points, then obtain a current written offer.

Sources and method

• Measurement Lab: understanding speed-test results ↗
• CRTC: internet performance and reliability ↗
• CRTC: universal service objective ↗

Measurement Lab explains that test results depend on the client, server, network path and measurement method. The CRTC separates service availability objectives from an individual test result and directs readers to consider performance and reliability evidence. This guide applies those principles to a repeatable troubleshooting sequence; it does not publish unverified provider benchmarks or diagnose a provider fault.

Run another speed test →