The curious case of Three's dwindling 5G coverage

Every few months, Ofcom, the regulator of the UK's telecom market, issues a Connected Nations report on the progress of England, Scotland, Wales and Northern Ireland (in no order of importance) toward the nirvana of ubiquitous full-fiber networks and gapless 5G. Charts and data typically show steady improvements in these coverage metrics. But something odd happened in the recent spring update. The outdoor 5G coverage offered by Three, the smallest of the country's four mobile network operators, went down.

It was not a slight decrease, either, but a jarring drop of 11 percentage points, to 67%, between September 2023 and January 2024. If the UK population were spread evenly across the land, like marmite across a slice of toast, this would remove Three's 5G coverage from nearly 7.4 million people. Since May 2022, no UK network, Three's included, has experienced as much as a percentage point dip between quarters. If nothing else, this is unusual.

What, then, is going on? The explanation provided by Three is all to do with a technology called dynamic spectrum sharing (DSS), which – as it basically says on the tin – allows frequencies to be divided between 4G and 5G networks as circumstances demand. Because DSS tends to rely on the lower frequency bands used with 4G, it is good for blanketing vast swathes of land. But it has led to some telco disappointment regarding capacity. "Some operators are taking DSS, which is effectively giving you a 5G symbol but 4G performance," said Nick Read, Vodafone's former CEO, after his own dismal encounter with it several years ago.

Three appears to agree with Read's assessment after some DSS experimentation. It is now removing the DSS products it installed and says that is why Ofcom's numbers have changed so dramatically. Three also points out that it has never tried to market DSS as a 5G technology to its customers.

In the company's own words: "We at Three only publicize our C-Band 5G coverage as this delivers the best non-standalone 5G experience. DSS 5G is a shared spectrum technology that provides speeds similar to 4G and is in the process of being switched off, so we have never promoted this coverage metric to our customers. We still have the largest C-Band coverage delivering the best (non-standalone) 5G speeds to Three UK customers."

The C-Band here refers to spectrum in and around the 3.5GHz range, where Three is extremely well resourced after takeovers and government auctions. Higher ranges come with more spectrum but offer worse coverage, partly because signals in these bands struggle to penetrate walls and other obstacles.

DSS dismay

The distinction Three makes between C-Band and DSS would certainly explain why its own coverage figures differ so markedly from Ofcom's. During a Glasgow press conference in January, Iain Milligan, Three's chief network officer, put 5G coverage at 62% and said Three had stopped expanding it in late 2022. "The big reason for that is the cost of deploying 5G is huge, and there are lots of other things to be focusing on," he said at the time.

Ofcom's data roughly tallies with this claim. According to that, Three's coverage soared from more than 40% in May 2022 to just less than 60% in September that year and then stayed at about the same level until April 2023. By September 2023, it had rocketed to nearly 80%, an increase Three would presumably attribute to short-lived experimentation with DSS technology.

It was obviously expecting more from DSS and reckoned the 5G experience was hardly different from what a user would see on 4G. Sites where DSS was deployed are accordingly being reconfigured to use Three's 2100MHz spectrum for nothing but 5G. The cost of all this to Three remains unclear, although extending C-Band coverage would undoubtedly have been more expensive.

All this has led to some confusion in analyst circles, and Three's DSS rollout has evidently not helped its case with UK authorities that are currently reviewing a proposed merger with Vodafone. The telcos insist they will be in a stronger position to invest in 5G after they combine. But in its detailed "Phase 1" review of those plans, published on April 29, the Competition and Markets Authority (CMA) reckoned 5G rollout was happening quickly enough without a merger. Three, it said, "has the most extensive outside premises coverage of 78%."

In contrast, Three's senior management is currently presenting the company as an unprofitable, sub-scale telco that will not be able to compete against BT and Virgin Media O2, or generate a decent return on capital employed, without a merger. But one of its problems has little to do with scale. Unlike other telcos, which would typically deploy multiple generations of mobile technology on the same vendor's platform, Three has more suppliers than generations. That is partly because it is still shifting its 5G network from Huawei to Ericsson after a government ban on Chinese companies was imposed. All this has led to some complexity, as Milligan recently explained.

"One of the challenges we have that's a part of the project we're working on for open RAN is we have quite a mixed-vendor radio estate at the moment," he said at the FutureNet World event held recently in London. "We still have 3G Nokia equipment, we have first-generation Samsung 4G sites, we have Huawei 4G, 5G, we have Ericsson 4G, 5G as well." In cities like Glasgow, where all four vendors are present, Milligan is investigating whether open RAN might allow the company to use one software platform across all those different radios and be more efficient.

The current circumstances must be a nightmare for Three's marketing department. Make the company sound way ahead of rivals and it risks undermining the pro-merger message to UK authorities about Three's investment plight. Depict Three as a laggard – hardly what marketing is all about anyway – and customers could defect. With the CMA now into its "Phase 2" investigation, Three is walking a tightrope.