How the FCC Measures Home Broadband Speeds

When a news story says internet providers delivered some percent of advertised speed, it usually comes from this program. Knowing how the measurement is made helps you judge your own speed test. This guide uses the FCC's Thirteenth Measuring Broadband America report, which covers data collected in September and October 2022.

Who is measured, and how?

The report calls its sample the MBA “panel.” Thousands of volunteers are drawn from the subscriber bases of participating providers. Each is sent a whitebox, which the report describes as pre-installed software on an off-the-shelf router that measures thirteen broadband performance metrics, including download speed, upload speed and latency. Because the box sits on the customer's own line, the program measures what the service delivers to a home, not what a lab shows.

The sample aims to include the tiers that make up the top 80% of each provider's subscribers, so uncommon plans may be left out. That is a design limit worth remembering when you read a headline.

Why a median, and why the evening?

For each whitebox the program takes the mean of all measurements, then for each speed tier it takes the median of those means. The report defines the median as the value separating the top half of a set from the lower half, “the middle (i.e., most typical) value.” Medians are less swayed by a few unusually good or poor results.

The report notes that peak operating hours have stayed the same as before the pandemic, 7 p.m. to 11 p.m. local time, even though more people work from home. Evening is when neighbors stream, game and call at the same time, so it is the fair test of a shared network.

What is the 80/80 measure?

The program also reports consistency. The 80/80 metric measures “the minimum percentage of the advertised speed experienced by at least 80% of subscribers for at least 80% of the time over peak periods.” A provider can have a good median and a poor 80/80 if results swing a lot from night to night.

In the Thirteenth Report, 12 provider and technology configurations were evaluated. The advertised download speed reported was 467 Mbps, as a weighted average. Of the major providers tested, eight measured 100% or better of advertised download speed in the peak hours, and the four others delivered between 86% and 90%. By the 80/80 measure, six of the 12 were at or above their advertised speed.

What is working latency?

Latency is usually measured on an idle line. The report also gives latency under load, which it calls working latency, meaning the average delay while the service is saturated. As expected, those results are higher than idle results. This is the same effect that makes a video call stutter when someone uploads a large file, covered in our bufferbloat guide.

How do you use this for your own test?

Borrow the program's habits. Test with a cable to the router, not Wi-Fi. Test at several times, including 7 p.m. to 11 p.m. Take several readings and look at the middle one. Write down the advertised speed next to each result. Our speed test guide shows the steps, and what up to means explains why the advertised number is a ceiling.

For High Desert, the plans are described as up to 500, up to 800 and up to 1 Gig down, with up to 200 Mbps up. If your evening results look low, call 775-575-9593 any day from 9 to 5 with your readings. If you are only weighing the service, send your address through the check form and a person will look at the path from the nearest tower and call you back.