For the second year in a row, Bell is Canada's fastest mobile network. But tough competition from Telus means Bell needs to stay on its toes.
With pandemic travel restrictions lifted after a rough 2020, we were thrilled to get on the road and test wireless networks throughout Canada this year. We drove through more than 40 cities and towns, as well as rural areas in eight provinces. Our quest: to find the performance gaps among carriers, between urban and rural regions, and in 4G and 5G download speeds.
We found a Canada divided. While urban areas are getting world-class speeds and rock-solid network reliability, rural areas struggle with much lower speeds and less reliable networks. And while Bell, Sasktel, and Telus have the airwaves to rule the 2020s, Freedom and Videotron—and, to a lesser extent, Rogers—don't have the spectrum resources to truly compete on performance.
Last year, Bell dominated our results, winning in 12 out of 20 cities. This year, of our top 20 cities, Bell and Telus each won eight, Rogers won two, and there were two ties between Bell and Telus. Our nationwide results are population-weighted—one Canadian, one phone, one vote—and Bell's wins in Toronto and Montreal pushed it over the edge this year. (Together, those metro areas are 42% of our overall score.) To see who won in your city or province, check out our full results below.
Overall, speeds stayed steady from 2020 to 2021, primarily because of a jump in network traffic in Canada. Speed comes from airwave capacity—if less data is traveling over a network, there’s room for it to travel faster—and Canadians are using far more gigabytes per month than they used to, thanks to newly available unlimited plans. But even when Canada's networks are relatively congested, they’re still faster than those found in the US. (Check out Fastest Mobile Networks 2021 for more details.) There’s plenty of room on this information superhighway.
Prefer free Wi-Fi with a side of coffee to 4G? See the Fastest Tim Hortons locations in Canada.
The next year promises to bring more dramatic change than 2021 did. Carriers recently bought a swath of 3,500MHz airwaves known as C-band or n77. This will dramatically speed up 5G in urban and suburban areas, though people who live near airports will largely be left out of network improvements.
In addition, Shaw, which runs Freedom Mobile, wants to sell itself to Rogers. So 2022 could usher in a new set of carriers and a jump in performance on new airwaves.
The Promise of 5G: Realized at Last?
Last year, we could only access 5G networks in half a dozen cities. This year, we saw 5G on at least one carrier in every major city; it was only absent in small towns. Here's where we found—and didn't find—5G.
5G is making a difference in Canada right now, but that difference is obscured by some smoke and mirrors.
Every major city we tested now has 5G from at least one of Bell, Rogers, and Telus, and we found distinct differences between 4G and 5G performance. In our most apples-to-apples test, where we visited the same Toronto locations with a Telus phone in 4G mode and then in 5G mode, we saw a 25% improvement in performance from 4G to 5G.
Our other comparisons weren't quite as scientific, because we were comparing 4G sections of cities to 5G sections of the same cities, but looking at four other locations—Halifax, Kitchener, Moncton, and Vancouver—we saw between 7% and 69% improvement in performance. (There’s one anomaly, where 5G areas of Halifax with Rogers showed lower performance than 4G areas, but that’s probably down to the different locations tested.)
In general, when Bell and Telus went head to head in a city, the winner was usually the carrier on which our phones more frequently attached to 5G. The speed improvement of 5G made a substantial difference in the consumer experience, which is the experience we care about the most.
Why does 5G matter? If you think of a network as a road that your data travels on, you might expect 5G to be a wide highway with lots of room on it. But in most cities, Bell and Telus are only using 10MHz of airwaves for 5G, out of a total of between 35MHz and 90MHz. Rogers uses 10 to 20MHz for 5G. That's a relatively small fraction of their overall airwaves, and it’s on channel sizes and frequencies where 4G would work just fine.
5G is about 30% more efficient than 4G, which accounts for a bit of the boost, but what we're probably really seeing here is that the 5G networks are relatively unloaded—there's less traffic on the road. According to Statista, 5G phones only have a 17% market share in North America. Four out of five phones aren't using the new airwaves at all, so the ones that can use them are getting big boosts.
If there's a lane with fewer cars in it, you want to be in that lane, so now's a good time to get a 5G phone that's compatible with C-band (which most of them are). You'll get a bit of a boost right away, and when the 3,500MHz 5G airwaves come online in 2022 or 2023, your data speed will jump even more.
The Role of Latency
When I sat down with carriers earlier this summer, they suggested that low latency (the time it takes for data to pass from one point on a network to another, usually measured in milliseconds) was a big advantage for 5G networks. Bell gave me two great new uses for 5G: sports fans panning around dozens of cameras at Montreal's Bell Centre, and Toronto startup Tiny Mile delivering food using remote-controlled delivery bots. The first one needs low latency so the cameras respond quickly to your commands; the second one needs it so the robots can react to obstacles such as erratically moving people. (From a robot’s perspective, all people move erratically.)
Latencies below 11ms aren't widespread yet. Rogers had by far the most latencies of 10ms or below, showing connections with latencies of 7 to 10ms in Montreal, Toronto, and Vancouver.
Rogers' particularly strong position on latency shows a great way it's speeding up its network for consumers despite lacking the spectrum to deliver massive download speeds. For web browsing, latency is the gumminess of a network—that dragging feeling as things start to load. Lowering latency makes your basic internet experience feel sprightlier and more responsive, and for Rogers customers in Toronto and BC, that can be very compelling. The only reason Rogers didn't do as well on our overall national latency measure is its poorer performance in other provinces.
Next year, when we upgrade our testing software, we'll be able to look more closely at the effect of latency on the overall browsing experience.
Leaving Smaller Carriers Behind
What we're seeing in Canada right now is just the tip of the 5G iceberg. The technology really comes into its own when applied to large contiguous blocks of spectrum, 40MHz or larger. The bigger the block, the better the 5G.
We took close looks at sample locations in each of our test cities to see how much spectrum our phones were using for their LTE (4G) connections. This doesn’t measure how much spectrum the carriers own in each city, but how much our phones were actively using. Each carrier still sets aside some airwaves for 3G, for instance, and we wouldn’t have measured that.
On average, our Bell and Telus phones used approximately 75MHz of spectrum each, our Rogers phone used 55MHz, our Videotron phone used 25MHz, and our Freedom phone used 20MHz.
Bell, Rogers, and Telus all have spectrum blocks big enough to grow into 5G. Shaw does not. Even in the one city where we found Freedom "5G" (in London), it was using only 35MHz total between 4G and 5G, half of the other major carriers' allotments. For Freedom and Videotron to improve their performance, they need more megahertz of spectrum in each city they cover. But their existing coverage areas largely don’t overlap, and though Videotron has bought new spectrum in some Freedom cities like Toronto and Edmonton, it hasn’t bought enough in each of those cities to match the larger carriers. Even if the two of them merge to create a national carrier, it would still be at a disadvantage to Bell, Rogers, and Telus, who would each have much more spectrum in major cities.
The gap will only increase when the new 3,500MHz spectrum comes online. The Bell-Telus alliance has 90MHz in Montreal, 90MHz in Halifax, 120MHz in Ottawa, and 130MHz in Edmonton. Though Videotron made some significant purchases (50MHz in Montreal and 30MHz in Ottawa, for example), no other carrier alone can keep up with the powerful pair.
This map shows the carriers’ purchases in the 3,500MHz band. It’s important to note that Xplornet, a major purchaser of airwaves, probably intends to use them for point-to-point wireless home internet, not mobile service. That’s great—Canadians need more home internet choices!—but it doesn’t factor into mobile competition.
Bringing Rural Canada to Life
Dead zones are real, and they're a problem. To urban residents, occasional spots without signal might be minor annoyances, but they can stretch over a considerable distance in rural areas and be a massive inconvenience to anyone who needs mobile phone or internet service.
Our tests generally showed 98 to 99% 4G network availability in Canadian cities. That doesn't mean you won't ever get dropped or blocked connections—every city had a few—but in general, the networks (other than Freedom) were reliable. That isn't the case in some rural regions, though.
Along rural roads between Ottawa and the Kawarthas in Ontario, between cities in New Brunswick, and in parts of rural Nova Scotia, we saw significant stretches where some or all of our phones lost signal completely. In many areas, all of the carriers would fail together, but I saw some differences from region to region:
In Ontario between Sudbury and North Bay and in the Kawarthas, Rogers was the most reliable.
In Saskatchewan, Rogers was less reliable than the other carriers.
Bell was a bit more reliable than the others in rural Nova Scotia.
Our regional results have all the details of our findings in each province's rural areas. Click the location links below to get the scoop.
The Best Wireless Plans in Canada
The biggest recent change in wireless plans happened at the end of 2019, when Rogers (and then Bell and Telus) introduced "unlimited" data. The Big Three's unlimited data plans aren't really unlimited—you get throttled after a while—but they get rid of overage charges. They're price-competitive with US mainline unlimited plans, according to WhistleOut's Stephen Clark, who tracks Canadian wireless prices.
The focus right now is on a Trudeau government promise to lower the prices on 2GB to 6GB plans by 25% relative to 2020 prices. That's led Bell's Virgin Plus to offer new “starter” 2GB plans for $37.50, 4GB for $41.25, and 6GB for $45 (all prices monthly); Fido and Koodo will probably match those prices within months.
But that 2GB to 6GB of data requirement is based on old data: specifically, StatCan's annual reports on wireless usage, which haven't been updated since the unlimited-plan revolution in 2019. At the end of 2019, StatCan said the average Canadian used 2.9GB of mobile data monthly that year, up from 2.5GB in 2018. On that basis, 2GB to 6GB looks like what Canadians want. But according to Cisco, Canadian data usage shot up to 5.2GB per line in 2021, and it's still well short of the 11.9GB-per-line average in the US. (Unlimited plans were launched in the US in 2016 to 2017, two years before Canada.) Canadians aren't natural data sippers; their cups were just kept half empty.
"Fear of data overage fees (currently at $13/100MB) drives the typical wireless consumer in Canada to avoid approaching their plan's data limit. I believe this has driven the market to advertise plans between 2GB and 6GB as sufficient for most consumers," Clark tells me.
Those new lower-cost Virgin "starter" plans have a poison pill attached, Clark points out. Unlike more expensive plans, they are bring-your-own-phone only. They don't allow financing a phone on a monthly payment plan, which is how most Canadians still prefer to purchase phones.
While prices are slowly getting lower and data buckets are getting larger in Canada, there's still nothing in Canada like the US' vibrant MVNO (virtual operator) market, which offers extremely low-cost single-line plans. The best-known US MVNO in Canada is Mint Mobile, because spokesman Ryan Reynolds is a major Canadian celebrity, but there are a dozen more like it.
"After accounting for the exchange rate, the average price for a talk and text plan (under 2GB) [in the US] is $17 USD, or ~$22 CAD. The cheapest talk and text plan in Canada is $22.50," Clark says.
This creates a real divide in Canadian pricing. While those who can pay at least $65 per line per month on a family plan are finally getting what they should, there’s very little competition for folks who can only afford $30 or less.
For the most up-to-date details on wireless plans, go to Clark's work at WhistleOut. Clark constantly tracks and updates plan prices throughout the year, whereas I only write this piece annually.
The Future: Two Grand Alliances
Rogers used to be Canada's leading wireless carrier, and when I started this project nine years ago, it won the first award. But Bell and Telus banding together created a spectrum juggernaut that Rogers couldn't match. Bell and Telus have traded off wins nearly every year of our project, based on network optimization, quirks of the unusual networks in Manitoba and Saskatchewan, and how each carrier’s handsets performed.
The fact is, Rogers can't catch up alone. Looking at recent purchases in the 3,500MHz band, in many major cities, Bell and Telus will maintain or extend their spectrum lead.
Rogers banding together with Shaw and Videotron could make the battle fairer. In Toronto, for instance, Bell and Telus have 165MHz after the 3,500MHz auction. Rogers has 140, but adding Videotron’s 20MHz and Shaw’s 10MHz could give the three of them a joint boost. In a world where your competitor is running on 165MHz, having 20MHz or even 50MHz to yourself is not a competitive offering at all. You need to be part of a larger team.
That suggests the future in Canada is two shared mega-networks. Two grand alliances of relatively equal performance, where the members of each alliance differentiate on customer service, core network reliability, multi-service bundling (cable, cloud gaming, US roaming, streaming services), and cultural affinities. There’s plenty of room to compete beyond radio networks.
Videotron has an existing, limited network-sharing agreement with Rogers in Quebec that has been going on since 2013. Videotron also previously discussed network sharing with Shaw, and analysts have said (according to The Globe and Mail) that Rogers may have to spin off Freedom Mobile to get the merger approved. Videotron has publicly said it wants Freedom Mobile. So all of the bones are there for this deal, except for a more comprehensive Videotron-Rogers sharing agreement.
Before you blanch at the idea of five or six competitors being reduced to two, that's what the situation is in Saskatchewan right now—the two radio networks are Sasktel/Bell/Telus and Rogers—and Sasktel offers the best unlimited plan in the country, according to WhistleOut. Two competitors can still be enough competition to keep prices low and offerings attractive.
There will have to be some sort of regulation to make sure that companies compete while they cooperate. But it's not impossible. One huge positive step the government could take would be to mandate at-cost access to virtual carriers such as Mint Mobile, as those MVNOs have played a major role in providing affordable mobile service in the US.
In 2017, I suggested—over carriers' protests—that they had sufficient capacity for unlimited data plans. I was right, and now they offer unlimited data plans. The Bell and Telus partnership has created a unique spectrum and performance giant in Canada. It looks like Bell optimized its core network and 5G availability in some of Canada’s top cities a bit better than Telus did this year, leading to its narrow overall win.
Unlike in the 4G era, 5G needs big spectrum blocks to show its strength. To keep competition flowing, Canada doesn't need more small, less-competitive networks. Vive la grande alliance.
Our Testing Methodology
We sent our drivers throughout Canada armed with Samsung Galaxy S21 phones using special, non-throttling SIM cards provided by the wireless carriers. The phones run custom testing software based on the Ookla Speedtest application. (Editors’ Note: Ookla is owned by Ziff Davis, PCMag's parent company.)
The software runs tests every two minutes, staggered so that carriers that share network elements aren't testing at the same time (Bell and Telus are staggered, for instance). Even-numbered tests ran to the nearest server in Ookla's network. Odd-numbered tests went to a server we ran in either Toronto (for tests east of Winnipeg) or near Seattle (for tests in Winnipeg or west) to simulate downloading data from a major cloud services provider. Generally, we didn't find a noticeable difference in speed between the servers, though we typically saw longer latency when going to our server rather than on-network servers.
In each city, we stopped at multiple locations for at least 15 minutes each, with the number of locations roughly proportional to the size of the metro area (though when our testers got enthusiastic and made more stops, we allowed it). Every three stops, we rebooted the phones and traded their positions.
This year we were under an unusual amount of scrutiny from both Bell and Telus, who asserted that because of their shared radio network in many areas, their phones should always be attached to the same network in the same ways. That's true in a lab, but the world is not a lab.
This is our ninth year of testing, and every year, the Bell and Telus phones have behaved slightly differently from each other, choosing different spectrum bands or attaching to different networks. That is a real-world scenario, and that's what we're testing. We made sure the phones were always on a level playing field in terms of when and how they were positioned, rebooted, and refreshed. Under that scenario, the Bell and Telus phones would sometimes make different choices, and those choices are reflected in our results.
Here’s how the various components of our speed score are allotted:
The "downloads over 25Mbps" and "uploads over 5Mbps" measurements fold in general network reliability, as a dead zone would record as not being an upload or download meeting the requirements.
We mention how frequently we saw 5G on our charts, but it doesn't factor into the overall score. Whether a phone has a 4G or a 5G icon on it is technically interesting, but if a carrier's 4G network is faster than another carrier's 5G network, we value speed over network type. What people really want is a consistent broadband experience, so our overall scores reflect that.
While we’ve marked Bell as Canada’s fastest mobile network this year, some other reports have preferred Telus. With their performance so close, this can come down to methodology. We use same-time, same-place, same-device-model drive-tested measurements with server selection split between nearby and remote servers. Crowdsourced reports such as Ookla and OpenSignal use a broader range of devices nationwide; however, if one carrier has a more rural user base than another, or one carrier’s users have generally older devices than another, that would penalize that carrier’s nationwide score on crowdsourced tests where it wouldn’t suffer on ours. Some of our testing involves routing to relatively distant AWS servers; if another test only uses nearby servers, it wouldn’t check issues in carriers’ core networks that may differentiate between them.
We also tested during a specific time frame, and networks are constantly changing. If another test used a different time frame before carriers had made core network improvements, it would have different results. Ookla’s most recent tests are based on the full third quarter of 2021, and we know (because the carriers told us) that Bell did core network improvements in Manitoba and Telus did so in New Brunswick during that period, for example.
5G performance also played a role this year. If a different test got more or fewer 5G samples from a particular carrier than we did (because of its geographic or device distribution), it may come up with different results. We stand by our results, but we also respect our competitors and partners.
Looking Ahead to Next Year's Tests
Next year, we are completely disrupting how we test networks. We’ll be using new software, Solutelia’s WIND solution, which will let us test elements like web page downloads, streaming, and voice calling. We’re already excited. In the meantime, click the location links below to see who wins where you live or work.
(Editors' Note: Solutelia, a brand under Ookla, is owned by PCMag's parent company, Ziff Davis.)