Everyone has experienced the frustration of calling a company and being passed around from department to department. One UK telecommunications company eliminated 1.3 million of these ‘transfers’ in a single year and reduced complaints to regulatory bodies by more than half.

Many of you have likely felt, ‘I want to improve the quality of our phone support, but I don’t know how to change the structure of our call center.’

When you call for support, being told ‘I will connect you to another department’ multiple times and having to repeat your explanation is a deep-seated issue faced by many companies across all industries. It not only wastes the customer’s time but also places a significant burden on the staff handling the calls.

In fact, there is a telecommunications company overseas that has almost entirely eliminated this ‘passing around’ using AI. It is the major UK telecommunications company, Virgin Media O2.

Today, we will look at the results of the company’s AI-driven call routing changes and the ‘next move’ they have implemented on top of those achievements. We will also organize some practical tips that Japanese companies can use as a reference.

Overview of the Case

Virgin Media O2 is a major telecommunications company providing fixed and mobile communication services in the UK, established as a joint venture between the US/European telecom giant Liberty Global and the Spanish telecom giant Telefónica.

In January 2026, the company announced that it had fully deployed a system using Natural Language Understanding (NLU) technology that connects customers directly to the appropriate department simply by having them state the reason for their inquiry at the start of the call. As a result, they reduced transfers between agents by 1.3 million in 2025, saving customers over 400,000 hours (approximately 24 million minutes). They also reported that complaints to the UK communications regulator, Ofcom, have decreased by more than 50% over the last 12 months.

The source of this information, in addition to Virgin Media O2’s official announcement, includes independent reports on the same figures by UK telecommunications industry media outlets Telecompaper and ISPreview.

Then, on September 24, 2026, the company announced its next move. They revealed a partnership with the US AI company Sierra to introduce new voice AI agents. They have begun a phased rollout, initially limited to routine inquiries regarding broadband faults.

In-depth Analysis

◆ Virgin Media O2’s Progress in Detailed Figures

The figures confirmed to date are as follows:

Reduction in call transfers (2025 results): 1.3 million
Time saved by customers (2025 results): Over 400,000 hours (approx. 24 million minutes)
Change in number of complaints to Ofcom (results for the last 12 months): Decreased by over 50%
Number of cross-skilled support staff: 5,000
Number of specialized team members focused on complex inquiries: Over 500
Scope of the new voice AI agent (announced September 24, 2026): Initially limited to a small fraction of total call volume, specifically broadband troubleshooting.
Valuation of partner company Sierra (after funding in May 2026): $15 billion.

There is an important distinction to make here. The figures of 1.3 million calls, 400,000 hours, and a 50% reduction in complaints are achievements that have already been realized. On the other hand, Sierra’s voice AI agent is a future initiative just announced in September 2026, and no figures demonstrating its effectiveness have been released yet. It is necessary to distinguish between these two stories—the company that halved complaints with AI, and the company that has further expanded its use of AI—as achievements versus plans.

◆ Why were they able to reduce complaints so much?

Think about it. When you call customer service, what is more frustrating than being kept on hold for minutes? Isn’t it the experience of “having explained everything, only to be transferred to a different department and forced to repeat the same story”?

The system introduced by Virgin Media O2 focused precisely on eliminating this “repetition.” When a customer states the reason for their inquiry at the beginning of the call, the AI understands the content and routes them to the appropriate department before they are connected to an agent. This structure reduces the burden on the customer by eliminating the need to re-explain their issue every time the responding agent changes, and it also makes it harder for complaints to arise in the first place.

Another reason is that this was not the company’s “first time” doing this. Since July 2025, the company has also been deploying an internal tool called “Lumi AI,” which provides agents with real-time access to past interaction history during customer calls. This new AI voice agent is an extension of these multi-year customer service transformation programs. It appears that the results are a product of steady progress rather than trying to change everything at once.

◆ Tips for application to Japanese companies

“Could we do something similar at our call center?” For those who thought this, here are some tips drawn from this case study.

Take inventory of one “department prone to call-shuffling”: First, write down tasks where the customer has to explain things multiple times because the department is split up, such as troubleshooting, payment verification, or contract confirmation.
Start with the most routine inquiries, not everything: Virgin Media O2 also introduced its new voice AI agent by limiting it to “broadband troubleshooting,” where both the cause and the response method are easy to pattern. It is safer not to expand to all departments at once.
Design it so that time saved by AI is redirected to complex tasks: The company is leaving routine tasks to AI while maintaining a specialized team of over 500 people. The key to maintaining customer satisfaction seems to be a design where “people focus on difficult cases because of what the AI handles,” rather than “reducing staff because the AI handles it.”

◆ Points to be aware of

There is one point I would like to mention for the sake of fairness. The achievement of 1.3 million calls and a 50% reduction in complaints is due to a system that has already been in operation for nearly a year: call routing via NLU. The new Sierra voice AI agent announced in September 2026 is a “future initiative” added on top of that track record, and its results have not yet been shown. If you confuse the two and understand it as “complaints were halved thanks to the new AI agent,” it will be inconsistent with the reality.

Also, it is known that other measures besides AI, such as training programs in partnership with the Institute of Customer Service (a UK customer service industry body), were being promoted simultaneously, which contributed to the reduction in complaints. It is necessary to keep in mind that there are parts that cannot be attributed solely to the effects of AI.

Why is it not easily reproducible by other companies?

Looking only at the numbers, one might think, “Shouldn’t we just introduce the same system?” but in reality, there are several barriers.

The first is that this is not a one-off measure, but part of a multi-year customer service transformation program. It is the result of steps taken over time: “Lumi AI” in July 2025, call routing via NLU in January 2026, and then Sierra’s voice AI agent in September 2026.

The second is the scale of investment. The partner company, Sierra, is one of the leading AI startups in the US, valued at $15 billion after its funding in May 2026. The parent company, Liberty Global, signed this partnership as a three-year contract, looking toward deployment across approximately 80 million fixed and mobile connections throughout Europe. An investment decision of this scale cannot be imitated by a single company on its own.

The third is that it was introduced while maintaining a sufficient human safety net of a specialized team of over 500 people. It is possible that customer satisfaction is being maintained precisely because they have not cut the personnel handling complex issues, even while expanding the scope of what is left to AI.

In other words, the starting point should be to think about ‘at what stage can our company begin,’ rather than simply borrowing the numbers from this case study.

Summary

Virgin Media O2 used AI for call routing, reducing call transfers by 1.3 million and saving customers over 400,000 hours in 2025, while cutting complaints by more than 50%
In September 2026, building on that track record, they partnered with $15 billion AI company Sierra to begin introducing voice AI agents (though this is still an ‘upcoming’ initiative)
Distinguishing between ‘already proven results’ and ‘upcoming initiatives’ is the key to correctly interpreting international announcements
Why not start by listing just one ‘department where customers are frequently passed around’ in your own company’s inquiry support?

AI is not a magic wand, but simply reviewing ‘where you are making customers repeat their explanations’ one by one can significantly change the customer experience. I hope the journey of Virgin Media O2 serves as a helpful reference for your company’s considerations.

■ Other overseas AI use cases found today

1. Document distribution service Scribd is classifying over 400 million documents with Google Cloud’s ‘Gemini Enterprise’ This is a case study where Scribd, one of the world’s largest document and book distribution services, is using Google Cloud’s Gemini Enterprise to classify content for over 400 million documents and 12 billion pages uploaded by users. By moving away from the traditional method of building specialized detection models individually and switching to multimodal AI that can read images and text simultaneously, they claim to have reduced processing costs by 50% compared to using conversational AI. This is a concept that can be referenced by any company, regardless of industry, that handles large volumes of documents and images. (Source: Scribd, Inc. classifies millions of documents on Gemini Enterprise, Google Cloud Official, September 24, 2026)

2. Marketing company Growth Digital, operating in 13 Latin American countries, sped up proposal creation by 5x with 113 AI agents This is a case study where Growth Digital, a marketing company operating in 13 Latin American countries, built 113 AI agents in-house using Google Cloud’s BigQuery and Gemini Enterprise. They report that they have accelerated proposal creation speed by 5 times, reduced campaign reporting work by 80%, and kept financial processing error rates below 0.01%. This is particularly relevant for companies that repeat the same operations across multiple locations and languages. (Source: How growing Latin American midsize businesses are building in the AI era, Google Cloud Official, September 24, 2026)

3. Chilean logistics company KLog.co sped up freight shipping document processing by 10x This is a case study where Chilean logistics company KLog.co used Google Cloud’s Gemini Enterprise to automate freight tracking and shipping document creation. They claim to have reduced manual data entry errors by over 90% and increased document processing capacity by 10 times. This is an easy-to-understand application example for companies with logistics and trade-related operations that still rely on paper documents and manual input. (Source: How growing Latin American midsize businesses are building in the AI era, Google Cloud Official, September 24, 2026)

End of article

Sources:

AI helps Virgin Media O2 avoid over one million call transfers as it saves customers more than 400,000 hours of time on the phone (Virgin Media O2 Official, January 15, 2026)
Virgin Media O2 Says AI Saved UK Customers 400,000 Hours in Support Calls (ISPreview UK, January 2026)
Virgin Media O2 reduces customer service call transfers by 1.3 million (Telecompaper)
Liberty Global begins Sierra AI rollout at Virgin Media O2 (broadbandtvnews.com, September 24, 2026)
Virgin Media O2 launches customer service AI voice agent on selected calls to make support faster and easier than ever before (Virgin Media O2 Official, September 24, 2026)
Virgin Media O2 rolls out new customer service AI voice agent (Telecompaper, September 24, 2026)

▼ Read this series in one place: [Link to magazine] ▼ Updates here: [Link to X (formerly Twitter) account]

■ Request to follow: On this account, I introduce one AI use case from overseas companies every day in Japanese. There are actually many ways of using AI that have become standard overseas but are not yet widely known in Japan. The purpose of this account is to deliver these ‘know it and get a step ahead’ cases with specific numbers and sources, in a format that is easy to apply to practical work. If you don’t want to miss the next article, I would be happy if you could follow and wait for it.

#note #noteDailyUpdate #DailyUpdate #TryLiking #AIUtilization #AIUseCases #GenerativeAI #Business #BusinessUtilization #OverseasCases #WorkEfficiency #Productivity #WorkStyle #WorkStyleReform #Technology #AIBusiness #OperationalEfficiency #CustomerSupport #TelecomIndustry #CallCenter #VirginMediaO2