The U.S. is a land of big frontiers, and so its networks have a lot of ground to cover. According to OpenSignal's most recent data, U.S. operators have done an exceptional job at providing consistent mobile data coverage to their customers. U.S. LTE networks are now among the most extensive in the world, but in terms of speed the U.S. finds itself falling short. Drawing on measurements from 182,000 OpenSignal users, we compared the overall 3G and 4G performance of all four nationwide operators and how their LTE networks stacked up in the largest cities.
T-Mobile traditionally has fallen short of its larger rivals Verizon and AT&T in coverage, but it's a problem CEO John Legere has vowed to fix. Our most recent data shows he's making good on that promise. T-Mobile's LTE coverage increased to 81% in the fourth quarter. It now nearly matches AT&T in LTE availability and is closing the gap with Verizon.
New LTE-Advanced networks have started pushing the upper boundaries of LTE download speed beyond 30 Mbps, but the U.S. is not one of the countries making that innovation leap. In fact, the U.S. 4G average of 9.9 Mbps falls well short of the global download average of 13.5 Mbps. Though U.S. 4G networks continue to get faster, they're just not keeping up with the world's quickening LTE pace.
All four major operators saw improvements to their coverage footprints in the last quarter, the result being that overall U.S. 4G subscribers can now see an LTE signal 81% of the time. Crossing the 80% coverage barrier is something only seven other countries have achieved.
Verizon is still the operator to beat when it comes to network reliability, but T-Mobile is squaring off against the super-carrier in download speed. Nationally both operators are averaging 4G connections of 12 Mbps, and in a speed comparison in the 11 largest U.S. cities, T-Mobile just barely edged out Verizon. AT&T and Sprint hardly even factored in the contest.
|Data Sample Size||376,893,289|
|User Sample Size||181,927|
|Sample Period||Oct 1st - Dec 31st 2015|
|Coverage: 4G||Latency: 3G||Latency: 4G||Download Speed: 3G||Download Speed: 4G|
Our app continually runs tests to measure the real world experience users receive. Instead of relying on user-initiated or drive-test simulations, we are able to paint a holistic picture of network’s performance through our background tests and crowdsourcing techniques -- all the while protecting the privacy of our millions of active OpenSignal users. The app has been downloaded over 15 million times collecting billions of measurements.
This metric shows the average latency on each network on 3G connections. Latency, measured in milliseconds, is the delay data experiences as it travels between points in the network. A lower score in this metric is a sign of a more responsive network.
|Atlanta||Draw between Verizon / T-Mobile||16.9Mbps average|
|Houston||Draw between AT&T / Verizon / T-Mobile / Sprint||9.8Mbps average|
|San Francisco||Draw between Verizon / T-Mobile||15.9Mbps average|
|Washington||Draw between Verizon / T-Mobile||9.2Mbps average|
The U.S. is a bit of a 4G enigma. It was one of the first countries to launch LTE back in 2010, and in the intervening five years it’s built one of the most extensive mobile broadband infrastructures in the world. U.S. 4G subscribers can expect to connect to an LTE network 81% of the time. According to our upcoming State of LTE report, the U.S. ranks among the top 10 countries in the world in terms of coverage, which is all the more impressive considering its broad geography. Most of its peers in the top 10 tend to be small and densely packed countries like Singapore and South Korea. (Though T-Mobile and AT&T refer to their HSPA networks as 4G, we use the term 4G solely to describe LTE networks.)
But that same network robustness doesn’t apply to 4G capacity. The U.S. is falling far behind globally in LTE speed. As many other countries start providing consistent 20 Mbps or greater connections, the U.S. average is 9.9 Mbps. That puts it in the same league as Argentina, a country that launched its first LTE network a year ago. That’s not to say that U.S. operators are delivering a poor LTE experience to their customers — 10 Mbps is more than enough to power any app on a smartphone — but the U.S. clearly is no longer pushing mobile technology boundaries like it used to. Before we delve deeper into the reasons why, let’s take a closer look at the dynamics of the U.S. market.
In a year-end blog post, CEO John Legere predicted that T-Mobile would surprise the world with the gains it made in coverage in 2015. Last year it started a nationwide rollout of LTE on the 700 MHz airwaves — low frequency spectrum intended to boost its coverage both outside of cities and into the hard-to-reach interiors of buildings. According to our measurements, T-Mobile is delivering on Legere's boasts. In the fourth quarter, T-Mobile had an LTE time-coverage metric (*) of 81%, which means its 4G customers were able to see an LTE signal 81% of the time in our tests (for more information on time coverage see our methodology page). T-Mobile's recent improvements now put it a little more than a percentage point away from matching AT&T in LTE network availability and helped it close the bigger gap between itself and Verizon. Sprint brought up the rear with a time coverage of 70%.
In past reports, T-Mobile has bested all operators in speed, but in the fourth quarter T-Mobile and Verizon were almost evenly matched. T-Mobile's 12.3 Mbps LTE average just barely edged out Verizon’s average of 12 Mbps, making it a statistical tie (AT&T and Sprint didn’t even come close). We delved deeper into our data to compare LTE speeds in the eleven biggest metro areas and found Verizon and T-Mobile battling out just as intensely on the local level. T-Mobile won the contest in four markets and Verizon won it in three, while three cities were statistical draws between the two (the 11th market, Houston, was a statistical tie between all four nationwide operators). Verizon still has the advantage in coverage. It's king when it comes to reliability, scoring a time coverage metric of 87%.
One final metric we would like to explore is latency, which measures the time it takes for data to make a round trip through the network. Low latency is important to overall quality of experience on a smartphone — webpages start loading faster — but it's critical to voice over LTE (VoLTE) as a lower network reaction time means a better quality call. Verizon, AT&T and T-Mobile have all begun moving their voice traffic from 2G to LTE networks. Surprisingly, though, the operator with the lowest latency is the one major provider that currently has no VoLTE services, Sprint.
That transition to VoLTE could help U.S. operators solve their 4G capacity problems. The majority of voice calls still go over 2G networks, but as U.S. operators move that traffic over to LTE they can start shutting down 2G networks and use that spectrum for 4G. T-Mobile recently stated that 40% of all its voice calls are now VoLTE calls. One of the reasons T-Mobile has made such big speed gains in recent years is it has been gradually turning off its 2G networks and putting that spectrum to work in its LTE network.
The U.S. won't catch up with the rest of the world solely through cannibalizing its 2G networks though. Its problems are a bit deeper, and they're both political and technological. One of the main reasons why countries in Europe and Asia have been able to deploy such ultrafast networks is that they're tapping into new and larger blocks of spectrum that can support much more 4G capacity and much faster connections. Furthermore, many of these international operators have been quite aggressive in their pursuit of new technology. They're using LTE-Advanced techniques to tie multiple spectrum bands together, allowing them to boost individual device speeds even higher.
The last major 4G spectrum auction in the U.S. was in 2008, and in the last few years most U.S. operators have been adding to their holdings through acquisitions and trades. As for LTE-Advanced, its use in the U.S. has been fairly limited, in part because U.S. operators don't have big blocks of spectrum to tie together. Another problem is the success of U.S. operators in signing up 4G subscribers over the last five years. So many customers competing for limited capacity means that average connection speeds are driven down.
There's still hope that the U.S. can get back in the global LTE speed race. This year the Federal Communications Commission plans to auction off a major chunk of the broadcast TV airwaves for mobile broadband use. Depending on how much spectrum is released and when it becomes available (it may very well be reserved for future 5G networks), U.S. operators could use those frequencies to reach parity with the world's LTE speed demons. But until then, the U.S. may well have to be content with being the tortoise of the 4G world: slow but reliable.
(*) Editor’s note: In June of 2016, OpenSignal changed the name its time coverage metric to network availability. Availability measures the same thing as time coverage — the proportion of time users remain connected to a particular network — but we felt that availability was a better reflection of the metric’s definition. For more details see our methodology page.
OpenSignal data is collected from regular consumer smartphones and recorded under conditions of normal usage. As opposed to drive-test data, which simulates the typical user experience by using the same devices to measure network performance in a small number of locations, we take our measurements from millions of smartphones owned by normal people who have downloaded the OpenSignal app.
Those measurements are taken wherever users happen to be, whether indoors or out, in a city or in the countryside, representing performance the way users experience it. For more information on how we collect and analyze our data see our methodology page.
For this particular report, 376,893,289 datapoints were collected from 181,927 users during the period: Oct 1st - Dec 31st 2015
For every metric we've calculated the statistical confidence interval and plotted this on all of the graphs. When confidence intervals overlap for a certain metric we can't actually be sure which of the overlapping operators has the best performance.
For this reason some metrics have multiple operator winners when we've judged that the data is too close to call a victory.