Original article ISPreview UK:
Network testing firm Ookla, which collects data via their popular broadband Speedtest.net service, has published a new study that examines the “striking gap” between the home broadband speeds users are promised by ISPs and the speeds they actually experience – “the biggest factor behind this disconnect is the in-home Wi-Fi environment“.
The new study, which uses data collected during Q1 2025, looks at the percentage of households in a country achieving specific broadband speed levels, based on real-world measurements taken from users’ devices. But in order to do this it appears to manufacturer its own speed tiers by looking at the percentage of their panel that experience different speed levels, which isn’t necessarily an accurate reflection of the package class that end-users have actually taken from their ISP.
“These indicators categorise household-like Wi-Fi environments, which are approximated based on user device behaviour and network characteristics, by the maximum download speeds measured over home Wi-Fi. We look at three performance tiers: equal to or above 30Mbps, 100Mbps, and 250Mbps,” said Ookla.
For example, the study found that 92% of home broadband users in the UK achieved the maximum download speed on Wi-Fi in the 30Mbps+ profile, which drops to 45% for those in their 100Mbps+ profile and 24% for those deemed to be on 250Mbps+ tiers. The United Kingdom actually does reasonably well in this table.
The table shows that nearly all broadband users across the analysed markets achieved download speeds equal or above 30Mbps on their home Wi-Fi networks. But the proportion of users achieving speeds meeting the 100Mbps tier was significantly lower across all countries, “indicating that access to this experienced performance tier remains relatively limited and may be considered a premium level of service in most markets,” said Ookla.
In the UK, speeds of c.100Mbps are considered more of an entry-level “full fibre” FTTP tier, and can be taken by 88% of premises (definition of gigabit-capable broadband coverage), so we’d tend to disagree that it’s a “premium level of service“. The realty in the UK is that a lot of consumers are still on slower packages because they haven’t felt a need or been able to upgrade yet.
In fairness, Ookla does point out that, based on Ofcom’s data, 83% of UK fixed broadband users were on plans offering download speeds of at least 100Mbps as of July 2024, with less than half, 44%, subscribing to tiers of 300Mbps or more.
Ookla then compares their experienced speed tiers against the network take-up of high-speed infrastructure in each of the countries tested, which naturally finds that the “two are not directly correlated“. As above, take-up of full fibre is a big factor, as is the end-user’s choice of package when they do adopt it (i.e. we don’t all buy the fastest tiers available to us as they’re often the most expensive and might not be necessary).
Lest we also forget that WiFi performance will vary based on the quality of your router (WiFi standard, chipset, antennas etc.), the device the end-user is testing with (Smartphone vs desktop computer etc.), distance from the broadband router and how much load their local network is under at the time. Suffice to say that we’re not sure how useful this data is.
Speaking of WiFi technology, Ookla noted that 23% of user tests conducted in the UK were spent while connected to devices using the old Wi-Fi 4 standard, while 57% were on Wi-Fi 5 and just 20% were connected via Wi-Fi 6 (support for Wi-Fi 7 is still too small to even register). Naturally, the more modern the Wi-Fi standard, the faster your local speeds are likely to be; assuming you have a broadband connection that can maximise it.
In an ideal world, the way to conduct a study like this properly would require everybody to be running speedtests in a much more scientifically controlled environment, which could compare wired vs WiFi speeds against the actual packages consumers have chosen across a broad spectrum of networks/providers. But that’s quite tricky to do at scale. Otherwise, Ookla’s research, while interesting, doesn’t really tell us anything new.
