Ookla London Mobile Speeds Analysis Shows 5G Adoption Outrunning Capacity

Ookla’s London mobile speeds analysis places the capital last among 18 European capitals, down from seventh in 2021 despite broader 5G adoption. The findings challenge a familiar assumption: moving more people onto a newer network automatically improves their connections. According to reporting on the study, London’s demand has grown faster than the capacity available to serve it.

The reversal involves both a declining position and a specific performance setback. London’s reported median 5G download speed fell between 2021 and 2026, while Stockholm now leads the comparison for overall mobile download speed. Those measurements describe different things, and combining them would exaggerate the evidence.

For operators, City Hall, and businesses relying on mobile connections, the pressure is practical. Coverage establishes whether a network reaches a location; capacity determines how much traffic it can handle there. London’s early deployment advantage has given way to a harder test: keeping service usable when the working population arrives.

Ookla London Mobile Speeds Reveal an Early Lead That Did Not Last

The headline ranking comes from an analysis attributed to Luke Kehoe, Ookla’s lead industry analyst. Reporting on the findings describes London’s fall from seventh among 18 European capitals in 2021 to last in the latest comparison.

London’s early 5G advantage did not translate into a lasting performance advantage.

The reported third-quarter 2026 median mobile download speed was 109.43 megabits per second, a measure of how much data a connection transfers each second. Stockholm recorded 332 Mbps. The European capital comparison also placed London last for median upload speed, at 11.40 Mbps.

A median is the midpoint of recorded results, with half above it and half below. It describes the center of the sample rather than the experience of every subscriber.

That distinction matters for anyone interpreting the ranking as a verdict on their own connection. A person working near an uncongested site might experience good service. Another person nearby, using a different operator or standing inside a building, might struggle.

The historical comparison also requires care. Reporting puts London’s 5G-only median download speed at 200 Mbps in the second quarter of 2021 and 133 Mbps in the third quarter of 2026. Those figures should not be compared directly with the latest all-mobile median.

The narrower conclusion is still consequential. The study describes declining performance within the newer technology alongside a worsening position across the broader mobile market.

Rankings alone cannot tell that whole story. A city can move down a league table while its connections improve, provided competitors improve faster. Here, the reported 5G figures add evidence of a setback beyond relative position.

For readers, the useful question is therefore broader than whether London occupies the bottom row. It is whether the network delivers better service as adoption expands.

Early deployment gives operators a head start, but that advantage has conditions. A relatively lightly used network can produce attractive results before a larger population begins sharing it.

Once adoption grows, performance depends increasingly on the equipment and infrastructure supporting those users. The original launch date becomes less informative than the capacity available at busy locations.

London’s experience illustrates that distinction without establishing that every early deployment follows the same trajectory. The ranking is a warning about maintaining an advantage, not evidence that adopting 5G was a mistake.

It also changes the benchmark for future announcements. More coverage remains useful, especially where service is missing. In an already connected business district, however, additional availability does not answer whether existing users receive enough capacity.

The event is significant because it shifts the burden of proof. Operators must show that expanded access produces a better working connection, rather than relying on the network generation displayed on a phone.

The Working Day Exposes the Gap Between Coverage and Capacity

A mobile network faces a different challenge when people concentrate around offices, transport hubs, shops, and public spaces. The number of residents alone cannot describe that demand.

Reporting on the study identifies a weekday performance penalty in central London. Weekday network congestion coincided with download speeds 27% below comparable weekend hours.

The pressure falls on the network’s ability to serve people where they gather, when they gather.

This is the central conflict between 5G adoption and 5G capacity. Adoption describes greater use of the technology. Capacity describes the traffic the network can support with its available resources.

A coverage map can establish an important baseline, but it cannot fully describe a lunchtime connection outside a crowded office building. Geographic reach and service under demand require different evidence.

For an operator, that creates a deployment problem with a local shape. Capacity added elsewhere does not automatically relieve a congested business district.

For employers, it creates a gap between nominal connectivity and dependable work. Staff may have mobile service available while still encountering delays during activities that require an immediate response.

Consider a hypothetical employee joining a video call while walking between meetings. The call involves sending and receiving data continuously, rather than simply downloading a document.

A connection that handles an occasional file comfortably might behave differently during an interactive session. A citywide download median does not capture every part of that experience.

The same distinction applies to cloud applications. A worker submitting a document, retrieving a shared presentation, or using an online assistant depends on the connection throughout the interaction.

These examples explain the stakes; they are not measured findings about particular applications in London. The study does not establish that any named software product fails across the capital.

The implications extend to business purchasing decisions. Comparing operators requires attention to the places and hours where employees actually work.

A national claim about reach can be accurate while remaining insufficient for that decision. A company choosing mobile service for a concentrated workforce needs evidence that matches its operating conditions.

Public authorities face a related challenge. Improving service involves the physical places where equipment can operate, as well as the commercial networks using those places.

That makes the forced response more demanding than revising a coverage target. Operators need to identify constrained locations, while authorities and property owners need workable arrangements for equipment.

The timeframe also matters. A configuration change and a new physical site are different interventions with different dependencies. Announcing a future improvement does not establish an immediate increase in usable capacity.

London’s weekday pattern therefore gives the ranking practical meaning. It points attention toward the mismatch between where demand concentrates and where the network can accommodate it.

More 5G Users Can Share the Same Bottleneck

The apparent contradiction disappears when adoption and capacity are treated separately. A larger share of users connecting through 5G does not establish that infrastructure has expanded proportionately.

A newer network label cannot compensate indefinitely for constrained physical resources.

A cell, the local coverage area served by radio equipment, distributes available resources among connected devices. More simultaneous activity can increase competition for those resources.

The radio connection is also part of a longer system. Backhaul, the connection carrying traffic from a mobile site into the wider network, must support what the radios deliver.

This creates several possible constraints. Adding radio capability without sufficient supporting connectivity can leave another part of the system under pressure.

Conversely, improving the supporting connection does not remove every limitation in the radio network. The practical task is to identify the constraint at each location.

That explanation is a general network mechanism, not a complete diagnosis of every London site. The published comparison cannot identify a single cause for every disappointing result.

It does explain why adoption figures require context. Moving customers onto 5G changes the technology carrying their traffic, but it does not make infrastructure requirements disappear.

Standalone 5G, which uses a dedicated 5G core network rather than relying on a 4G core, is another relevant distinction. Its availability should not be treated as a guarantee of unconstrained service.

Ofcom’s mobile experience research examines connection success, transfer times, responsiveness, and speed distributions. That broader approach reflects the limits of judging service through a single headline measure.

Different metrics illuminate different problems. Download speed concerns receiving data; upload speed concerns sending it. Latency measures the delay involved in exchanging data.

Latency under load measures responsiveness while a connection is busy. It helps distinguish a connection that transfers data quickly from one that also remains responsive during that transfer.

For interactive work, this distinction can matter as much as a headline download result. Waiting for a response interrupts an activity differently from waiting for a large file.

Operators can respond through different combinations of additional sites, equipment upgrades, and network changes. The appropriate combination depends on local conditions rather than a universal recipe.

The comparison between London and Stockholm supplies a useful reference point. It shows that the reported performance level is not an unavoidable outcome for every European capital.

It does not prove that Stockholm’s infrastructure arrangements can transfer directly to London. Buildings, deployment histories, demand patterns, and operating conditions differ.

The fair comparison asks which practical choices support better service and which barriers prevent those choices elsewhere. Simply naming a faster city leaves that question unanswered.

For the same reason, the study should not become a simplistic contest between network generations. The relevant conflict remains expanding use versus the resources needed to support it.

The promise of 5G depends on continued deployment after launch. London’s results suggest that the maintenance of capacity deserves as much scrutiny as the original rollout.

The Hard Part Is Keeping Sites Available and Adding New Ones

Capacity is partly a physical infrastructure problem. Operators need places for equipment, connections to those places, and arrangements that keep sites available.

According to mobile deployment reporting, around three-quarters of London applications for new street masts since 2019 were refused. The reporting also describes disruption when rooftop sites must close during redevelopment.

Adding capacity and preventing the loss of existing capacity belong in the same discussion.

A replacement site is not automatically available when an existing location disappears. The network can face a service gap while alternative arrangements are developed.

That creates a less visible problem than a delayed launch. An operator might be working to restore lost resources while customers expect every new investment to deliver an improvement.

Kehoe’s comments to The Register identify planning restrictions, including conservation-area constraints, as obstacles. His explanation places the physical deployment process alongside growing demand.

The operator perspective deserves consideration because deployment requires cooperation beyond the network company. Equipment cannot simply appear wherever a technical model identifies a need.

The public-interest perspective also matters. Planning decisions weigh mobile infrastructure alongside the use and appearance of streets and buildings.

Treating every objection as unreasonable would skip that responsibility. Treating connectivity as somebody else’s problem would ignore the service consequences.

The practical question is how to make those decisions predictable while accommodating the infrastructure a working city requires.

An operator’s responsibility does not end at identifying external barriers. It still needs to explain where its investment will improve service and how progress will be measured.

Likewise, a policy proposal should be assessed by what it changes in delivery. A supportive statement does not establish that equipment has entered service.

There is also a difference between preserving a location and increasing its capacity. Keeping an existing site operational prevents a setback; it does not necessarily meet growing demand.

Those goals can reinforce each other, but they should remain distinguishable in public reporting. Otherwise, restored service can be presented as an expansion without explaining the earlier loss.

Temporary deployments provide a concrete example of location-specific planning. Vodafone described event capacity measures, including a temporary mast near the start of the 2026 London Marathon.

That operator announcement illustrates how anticipated demand can prompt targeted action. It does not establish that temporary event arrangements solve recurring weekday congestion.

The comparison is useful because it makes the capacity problem tangible. Demand arrives in particular places, so the response must account for those places.

For London’s working population, the more demanding challenge is repetition. The network must accommodate ordinary busy periods, not only exceptional events with dedicated preparations.

Progress therefore requires both operational decisions and durable access to infrastructure. Neither a marketing campaign nor a planning reform alone demonstrates that the bottleneck has been removed.

The Ranking Is a Warning, Not a Verdict on Every Connection

The findings deserve attention, but they do not support every conclusion readers might draw from the headline. A comparative study is evidence within a defined scope.

The strongest reading preserves the warning while respecting what the measurements cannot establish.

The first limit concerns geography. Last among 18 selected European capitals does not mean last among every European capital or every major city worldwide.

The second concerns the sample. Speed tests describe recorded connections, and the composition of those connections matters when interpreting change over time.

Devices, locations, operators, and testing behavior can influence an observed distribution. A careful historical comparison needs to consider whether those factors changed.

That does not make crowdsourced measurements unhelpful. It means the methodology belongs alongside the result, especially when a ranking becomes a broad claim about urban infrastructure.

Ofcom has separately cautioned that year-to-year differences in its research can reflect factors beyond network performance. Its observations concern its own study, not a direct validation or rejection of Ookla’s analysis.

The third limit concerns aggregation. Combining operators into a citywide result can conceal differences relevant to an individual customer.

A strong result for one network does not erase a weak result elsewhere. A weak combined result also does not establish that switching operators will solve every user’s problem.

The fourth concerns causation. A weekday slowdown is consistent with demand pressure, but it does not isolate every contribution from equipment, site availability, or network configuration.

Likewise, planning refusals provide relevant context without proving that each refusal caused a measurable decline. Establishing that connection requires more detailed evidence.

The fifth concerns the historical baseline. The pandemic changed commuting patterns, making comparisons with 2021 particularly sensitive to where and when people used mobile networks.

That context does not erase the decline. It helps explain why an early performance advantage should not automatically be interpreted as sustainable under later demand.

There is also a gap between speed and usefulness. A slower connection can still handle many routine tasks, while a faster connection can disappoint during unstable or delayed interactions.

Calling London unusable would therefore go beyond the evidence. Calling the ranking irrelevant would dismiss a meaningful signal about competitive performance.

The London connectivity dashboard offers a complementary approach by examining download speed, upload speed, signal strength, and latency across locations.

Its value lies in moving from a citywide headline toward a more specific question: where does service fall short, and under what conditions?

For businesses and residents, that local view is the bridge between a reported ranking and an actionable understanding of their own connections.

Watch Busy-Hour Performance Before Celebrating Another Coverage Milestone

The next stage should be judged through evidence that tests the capacity explanation. Over the coming months, three signals deserve attention in a fixed order.

The clearest improvement would be better service during the periods that currently expose the network’s limits.

First, watch whether the weekday performance gap narrows in comparable locations and time periods. That would strengthen the case that users are receiving more usable capacity.

A better quiet-period result would still be welcome. However, it would offer weaker evidence that the working-day constraint has eased.

Comparability matters here. A different mix of locations or operators can change an aggregate without showing what happened at the original pressure points.

The useful follow-up therefore pairs the headline with a clear account of the sample. Readers need to know whether an improvement reflects better service under similar conditions.

Second, watch upload performance and responsiveness while connections are busy. A download improvement alone would leave important parts of the working experience unresolved.

If those measures improve together, the evidence would support a broader service gain. If they diverge, the result would call for a more specific diagnosis.

This is particularly relevant for people using mobile connections interactively. Sending work, participating in calls, and exchanging requests with cloud services involve more than receiving data.

Third, watch completed infrastructure changes at constrained locations. New equipment entering service and replacement sites operating provide stronger evidence than an announcement of intent.

Planning approvals are useful intermediate signals, but they are not equivalent to operating capacity. The follow-through matters.

If busy-hour performance improves alongside completed deployments, the capacity explanation gains support. If deployments advance without better service, other constraints deserve closer examination.

If performance improves without the expected physical changes, operators should explain what altered. Better configuration or other network work might be contributing.

None of these outcomes requires London to regain a particular ranking immediately. Other capitals will continue changing, so relative position can obscure local progress.

The more useful near-term standard is whether comparable users receive a better connection when demand is concentrated. That directly tests the problem identified by the analysis.

For anyone assessing a network announcement, start with the location and working conditions behind the claim. Compare busy-hour results, check upload performance, and ask whether responsiveness improves alongside downloads. Separate planned infrastructure from equipment already carrying traffic. The Ookla London mobile speeds findings offer a reason to demand that detail, rather than treating broader 5G availability as a complete answer. Businesses should evaluate service where employees actually rely on it, while residents should look beyond a citywide median when comparing their options. Read the next measurement release with those distinctions in mind: does it show more people reaching 5G, or a network serving those people better?