(Image: Bob Al-Greene)
In 2022, T-Mobile isn’t just America’s fastest network, it’s also the best.
For this year’s nationwide mobile network tests, we’ve changed everything. We drove more than 10,000 miles across the country with new software that tracks dropped calls and provides a better measure of reliability. That turned this year’s project into a search for America’s best mobile network—not just the fastest, as we’ve called this project in previous years. (For more, see “Our Testing Methodology” below.) We’ve also shifted the way we drive, using electric and hybrid cars for portions of our route. (Read more about the actual drive in the “Going Electric” section below.)
This time around, we hit 30 cities and six rural regions using the latest Samsung Galaxy S22+ phones to see how AT&T, T-Mobile, and Verizon compare. T-Mobile wins 18 of our 30 cities to Verizon’s eight and AT&T’s four.
Look purely at speed, and the difference is even more stark. AT&T’s entire advantage is in places where it has greater network reliability than T-Mobile. T-Mobile wins 19 cities on speed, to Verizon’s nine and AT&T’s two.
T-Mobile is also doing better than ever before in rural areas, although it gets docked slightly for still having more rural data dead zones outside the Northeast than the other two carriers do.
Why Rate Mobile Networks?
Mobile internet is the internet, now more than ever. According to Ericsson(Opens in a new window), between 2018 and 2022, mobile data traffic in North America quadrupled from 22 exabytes to 86 exabytes. Most of that growth is driven by the revolution in mobile video, which has risen from 60% to 70% of mobile data traffic. That’s almost 20 gigs of usage, per person, per month.
Meanwhile, there’s a lot of noise around 5G. The carriers haven’t made anything easy or clear. They conflate different types of 5G with different performance characteristics. They claim very broad coverage for forms of 5G that act just like 4G. Sometimes, they put icons on your screen that make it difficult to tell which “G” you’re on, or why you should care. Someone needs to cut through the marketing language and tell you what’s really available, and which phone you need to use with it.
A lot of ratings are out there, and they’re confusing because they’re largely opaque. We started the Fastest Mobile Networks project aiming to be as transparent as possible. Each year, we show you exactly where we go, which devices we use and why, and give you every component of our overall scores. We want you to have as much power here as possible when making a buying decision on a wireless carrier.
5G Continues to Improve
Yes, you need a 5G phone to hit these high speeds. On both T-Mobile and Verizon, the difference between 4G and 5G is becoming huge now, and it can be the difference between a stalled connection in a congested lane and speeding along in the HOV lane of mobile internet. (AT&T isn’t quite there yet, but it likely will be next year.)
To test, we’re using the Samsung Galaxy S22+ because it has shown the best network performance in our tests. With its Qualcomm X65 modem, the S22+ can hold onto weak signals longer than phones from previous years, and it supports all the new 5G bands carriers are implementing, including the frequencies AT&T is launching later this year.
Since we used new software this time around, our results aren’t directly comparable to those from previous years. But some trends are difficult to ignore. Both T-Mobile and Verizon see dramatically increased nationwide speeds this year as they expand their mid-band 5G networks. But AT&T’s speeds decline, as it has added very little capacity while data demands grew.
The speed results and dropped-call results appear completely decoupled. In general, all three carriers show very low rates of dropped calls in cities. But AT&T is noticeably more reliable on calling in rural areas—able to keep that little bit of voice connectivity going even if there isn’t massive data capacity for streaming and gaming.
We use people, not square miles, to determine our scores. Rural areas comprise 16% of our score, similar to the 14-20% of the US population classified as rural.
Bringing Networks Home
As mentioned, our mobile network measurements matter more than ever because there’s a new use for 5G: breaking your local cable company’s home internet monopoly.
Smartphones are ubiquitous in America now, and since 2010, we’ve been giving you the most transparent, consistent measurements to help you choose your carrier. But it’s completely legitimate to wonder why you need more than 100Mbps on your smartphone at all.
There are two major reasons. First, that average speed also tends to define the floor. You may not feel you need 100Mbps, but you probably always want at least 25Mbps, and those greater average speeds tend to link up with more consistently reliable minimums.
But also, both T-Mobile and Verizon are selling home internet service now. That’s where you really will want 100Mbps or more, and our results show where you can probably get it. It’s always good to ask your friends and neighbors how well the service works or to test drive a unit yourself. But some solid third-party scores can really help you decide whether it’s a service worth considering.
You can clearly see T-Mobile’s triumph in these charts of overall results collected by speed bins. In the cities, T-Mobile has by far the least connections under 50Mbps, and it really pulls away from the competition in connections between 300 to 600Mbps, which is the mid-band 5G zone. This shows where T-Mobile probably has sufficient capacity for its new $50-per-month home internet service, too. In rural areas, the effect is less pronounced, but it’s still there.
Looking at dropped calls in the cities, the results are all over the map. In rural areas, T-Mobile suffers because of two specific regions: the rural Northwest, going from northern California to Washington; and Georgia. In the Northeast, coverage is solid.
On parts of our drives, we were able to check whether our phones were showing 4G or 5G, and what kind of 5G. T-Mobile showed the best 5G coverage—and the best high-quality 5G coverage—in both cities and rural areas.
Focusing on the good stuff, mid-band 5G, which works and feels better than 4G, in rural areas T-Mobile showed “5G UC” 43% of the time, compared with Verizon’s 5G UW 9% and AT&T’s 5G+ 2%. And the same goes for cities. Across 19 cities where we felt we had sufficient data, we saw T-Mobile’s high-quality 5G UC 78% of the time, compared with Verizon’s 5G UW 20% of the time and AT&T’s 5G+ 7% of the time. This is all because T-Mobile had a head start, building its mid-band network from Sprint’s airwaves, while Verizon and AT&T had to wait for later spectrum auctions. Verizon and AT&T are racing to catch up, and they very well could—just not yet.
T-Mobile’s winning secret is mid-band 5G, which it calls “ultra capacity.” The math of wireless network performance can be pretty simple: Use more airwaves with more reach and you have more performance.
With its purchase of Sprint in 2020, T-Mobile nabbed a massive cache of spectrum that it has quickly repurposed for 5G. That spectrum can reach a few miles from a tower, in best-case circumstances, so it’s good for covering cities and suburbs.
T-Mobile’s mid-band is present not only in cities and suburbs, but in rural areas too. Many interstate corridors nationwide showed T-Mobile’s “UC” icon, which signifies mid-band connections, and they also showed mid-band speed in our testing.
The orange spots show where we saw T-Mobile’s “UC” network in our testing.
In a few Northeast cities where we were able to take samples, we found T-Mobile using 10-15MHz of low-band n71, and 40-100MHz of mid-band n41. Sometimes, T-Mobile would combine the two for between 50-110MHz of n41/n71.
Verizon and AT&T are closing the gap. Both of the carriers purchased mid-band in an auction ending in February 2021; Verizon started building its C-band network immediately, while AT&T has been largely waiting for equipment that will combine its C-band mid-band with some more spectrum it bought in early 2022.
We saw a significant amount of Verizon’s “UW” icon nationwide and less of AT&T’s “5G+.” Both icons can signify mid-band or the even faster high-band spectrum. AT&T had the most mid-band-like characteristics in Detroit, Texas, and southern Florida. But both AT&T and Verizon are using less mid-band than T-Mobile is right now. While T-Mobile uses up to 110MHz, AT&T generally uses 40MHz and Verizon 60MHz.
This all affects speed. Simply put, more mid-band is good. T-Mobile has the most of it right now, so it’s the fastest network; Verizon has the second-most, so it’s the second-fastest; and AT&T is likely to hustle to catch up over the next year.
What About Dish?
There’s a new player coming to town. Dish aims to be the fourth national carrier, and has launched in more than 120 cities and towns in June—in places as big as Raleigh-Durham, NC, and as small as Eagle, ID.
Its current service is $30 per month, with only one phone offered (a Samsung Galaxy S22), though the company promises more phones in the future. Early testers say performance on Dish’s network reaches about 200Mbps down, and the phones roam onto AT&T when they’re outside Dish’s limited coverage area. So to some extent, you can use AT&T’s results as a guideline for Dish speed and reliability until it has been built out further.
We found in our tests that AT&T provides a highly reliable network, and that advantage is magnified more in rural areas. In upstate New York, for example, where Dish intends to launch seven cities this month, AT&T coverage dominates (although T-Mobile provides much higher speeds where it has coverage.)
Dish launched too late for our tests, which ran in May, but we’ll have ongoing coverage of its progress throughout the year.
Our Testing Methodology
A lot has changed this year. The biggest shift is switching from the Ookla software we used from 2016 through 2021, to a new package called Ookla Wind. (Note: Ookla is owned by Ziff Davis, PCMag’s parent company.)
Wind gives us the ability to test phone calls as well as speed. During our 10,000-mile drive, our phones would run a speed test to the nearest Ookla server, followed by a 30-second phone call to a recorded message, then pause for a minute to cool down before repeating.
Wind makes the phones run hot, despite the cool-down periods. So we strapped Razer Chroma phone coolers to the back of each handset, which can bring down the temperatures by as much as 10 degrees Fahrenheit.
The version of Wind we use on our speed test phones can’t discern the difference between 4G and 5G networks. To do that, we created a routine that takes a screenshot of the status bar once a minute. That only worked some of the time, but often enough to provide broad conclusions. We also had a select few phones with a more advanced version of Wind that could detect different forms of 5G, which we used in a few cities.
As before, we tried to stop in at least 12 locations per city for at least 15 minutes each. Speed results for a city are the average of all locations, with tests run on the move given the weight of two stationary locations. Reliability results are simply the success percentage of all tests taken in the city.
Measuring dropped calls means our scoring has to change, too. There’s generally a very small difference between the carriers’ dropped-call percentages, but it’s meaningful to consumers. And there are huge differences between average speeds, but they aren’t as meaningful to many people as a small difference in reliability.
So our scoring is now rank-based. For each component of our score, we rank the carriers and assigned points based on rank. So the third-place carrier always receives the same amount of points no matter what the difference is between it and the first-place carrier. The effect is to magnify reliability, which is what we’re looking for.
This year, our score consists of:
Click through our city and rural area links below (or from the drop-down menu up top) to see the results where you live, work, or play.
This year, we’ve made a big shift in not only how we test networks in America, but how we drive around it. We worked with Hertz to rent a Tesla Model 3 Long Range for the longest leg of the trip, a 4,500-mile loop around the West Coast, to prove that it’s possible to take a truly epic road trip in an electric car.
The blue lines show our driver routes for testing.
Hertz is transforming its rental fleet with up to 160,000 Tesla and Polestar vehicles over the next few years, making it the first large-scale electric car rental operation. We talk to Hertz about why it’s making that shift. Our testers have some views as well. Veteran driver, Angela Moscaritolo, found the Tesla to be a luxurious experience, while first-time Best Mobile Networks driver, Rob Pegoraro, had some gripes. See how our drivers, who are well outside of the Tesla fan bubble, found their EV driving experiences.