Chief Operating Officer / Japan
Decisions that changed how our network behaves under strain
- Customer households
- 441,000
- Kumamoto, Kagoshima and Miyazaki prefectures, March 2026
- Homes passed now on fibre
- 81%
- 496,000 of 612,000; coaxial retirement due by March 2029
- Annual operating cost in my remit
- ¥31bn
- Network, field, contact centres and community channels; fiscal year to March 2026
Decisions & consequences
01Choosing the restoration order after the April 2016 earthquakes
In the early hours of 16 April 2016, the second Kumamoto earthquake cut service to 118,000 of our 418,000 customer households. A hub site near Mashiki lost power and its building was declared unsafe, and two of our main fibre routes were broken where roads had collapsed. We had a written list for hospitals and municipal disaster headquarters, and nothing beyond it. By dawn I had about forty crews available and roughly ten times that much work.
The decision was which customers came next. Our head of enterprise sales wanted the business parks restored first: about 1,900 business sites had contracted service levels, credits were payable after twenty-four hours and several large customers were already calling. I proposed instead to restore the community channel's disaster feed, the links to municipal offices and the 61 evacuation centres in our area, then welfare facilities, and only then to work by the number of households a crew-day could reconnect. Our finance team estimated the credits at over ¥300 million. The president backed my proposal on the second morning, after hearing both of us.
We gave up commercial goodwill we had spent years building. Crews rebuilt the community channel's playout over a temporary fibre route and a portable system, and it was back on air in 29 hours carrying municipal notices. Within six days every evacuation centre had free Wi-Fi and a television on the community channel. Crews from four operators outside Kyushu arrived under arrangements I negotiated by telephone.
Ninety per cent of affected households were reconnected within 17 days and every reachable home within 38. We paid ¥310 million in service credits, and two enterprise customers worth about ¥180 million a year moved to national carriers within the following year.
What changed was the method. Since 2017 a written restoration order has been agreed each spring with fourteen municipalities and approved by our board, with its likely cost stated, and the Minami-Kyushu Cable Mutual Aid Council now provides crews on pre-agreed terms. What I would do differently is twofold. We published fixed restoration dates for 31 areas and missed nine; ranges would have been more honest. And our visiting crews slept in their vans for the first week because nobody had arranged accommodation or fuel for them in advance. Both now sit in our standing preparations.
Moving whole neighbourhoods from coaxial cable to fibre
By 2017 much of Hanagiri's coaxial plant was more than twenty years old, fault rates were rising and spare parts were becoming hard to source. I was asked to write the replacement plan.
There were two credible options. Our consumer sales division proposed a demand-led programme: build fibre to an area, sell upgrades to households that wanted faster broadband at a higher tariff, and keep the coaxial network running for everyone else until about 2035. It would cost about ¥41 billion over ten years and generate upgrade revenue along the way. I proposed migrating whole node areas, moving every household to fibre at our cost with no change in tariff and switching off the coaxial plant behind them, completing by March 2029 for about ¥58 billion. The head of consumer sales opposed my option in writing; he expected to forgo much of the tariff uplift and disliked visiting customers who had asked for nothing. Our chief financial officer asked for stage gates every two years.
The board approved node-by-node migration with those gates. We gave up a faster return on capital and accepted running two networks for less time but at higher initial cost. Every household received a survey, an appointment, a set-top box exchange where needed and a second visit if anything went wrong, and neighbourhood associations were briefed before each area began.
By March 2026, 238 of 291 node areas had been migrated and fibre reached 496,000 of the 612,000 homes we pass. Fault tickets fell from 14.2 to 5.1 per thousand customers a month, field visits by 41% and plant electricity consumption by 28%. Annual churn among migrated households is 0.6 points lower than before. Broadband upgrade take-up has been 19%, against 27% in the sales case, and the forecast cost has risen to ¥61 billion, about 5% over plan.
The decision made the coaxial network's end date a commitment rather than a hope. What I would do differently is the start. In the first eleven node areas, 9% of households needed a second visit because of in-home wiring we had not surveyed, and we received 3,100 complaints in two years, mostly from older television-only customers about appointments. Surveys before every visit, and earlier work with neighbourhood associations, should have been in the plan from the first day.
Retainers for the contractors who keep the network standing
In 2022 our purchasing data on the 46 firms that carry out our field work showed cost per job, completion times and rework rates, but not how old the technicians were. When we asked the firms directly, the average was 51 and almost a quarter were over 60. Five firms had closed in three years, three because the owner retired without a successor. Together those closures described a field workforce leaving faster than anyone was joining, ahead of overtime limits for construction work that would apply from April 2024 and remove the old answer to a typhoon surge.
I proposed moving our eighteen largest firms, which carried about 71% of field volume, onto three-year agreements with a monthly retainer for each committed technician, unit rates raised by 12%, and conditions: one apprentice for every eight technicians, a jointly reviewed safety record and no subcontracting of our work beyond one tier. Our chief financial officer, Chiaki Morozumi, opposed the plan in writing because it added about ¥1.4 billion of fixed cost to an activity whose volume should fall as fibre replaced coaxial cable. Our head of purchasing wanted an open tender. Both memos went to the board with mine, and the board approved the plan subject to a three-year review against measures set in advance.
We gave up price pressure and flexibility, and the 28 smaller firms outside the retainers saw their share of our volume fall from 29% to 22%. Apprentices trained through the Hinoakari Technical Training Foundation.
By March 2026 technicians under 35 had risen from 9% to 17% of our contracted workforce, 164 apprentices had completed training and annual technician turnover had fallen from 14% to 8%. Average time to repair a reported fault fell from 31 hours to 19. After a typhoon in late August 2024, 41,000 affected households were restored within six days. Cost per completed job rose by about 7%.
The board now receives an annual account of how old and how deep our field workforce is. What I would do differently is the treatment of small firms. Two of them closed after losing volume, including the only firm based on our islands, and we now pay a premium to send crews by ferry. A pooled retainer that small firms could share should have been part of the design.
One master control room for seven community channels
Each of the seven stations that formed Hanagiri had kept its own community channel and its own master control room. By 2021 most of that playout equipment was at the end of its life, and replacing it in seven places would have cost about ¥2.6 billion. The channels are not a commercial afterthought. Municipalities rely on them to reach older residents with evacuation notices, and several have agreements with us covering disaster broadcasting.
I proposed one master control room in Kumamoto, with playout for all seven channels, remote control of five local studios and a direct link from every municipal office so that a town could put an evacuation notice on screen itself. The capital cost was about ¥1.1 billion. One outside director asked whether we should go further and outsource local production, which would have raised annual savings from about ¥290 million to ¥520 million by removing most of our sixty local reporters and crew. The mayors of two towns objected to the consolidation itself: they feared their disaster broadcasts would depend on a room an hour away.
I recommended consolidation without outsourcing, and the board agreed. We gave up about ¥230 million a year of further savings to keep reporters in their towns, and we merged the two smallest studios into neighbouring ones, which cost those towns a local studio. I met both mayors twice, demonstrated the direct insertion link in their own offices and agreed quarterly tests with every municipality.
The new master control went live in 2022. Annual operating cost fell by ¥290 million and local programme hours have risen 18%, because reporters freed from playout shifts now produce programmes. During heavy rain in July 2025, municipal offices inserted eleven evacuation notices directly, with a median of four minutes from decision to screen.
Disaster broadcasting is now tested quarterly and reported to our board each year. What I would do differently is the backup. To save money in the first year, we deferred the backup playout site in Kagoshima. In November 2023 a power fault at the Kumamoto site took all seven channels off air for 47 minutes. It happened on a calm evening, but a single point of failure in a disaster service should never have been accepted to bring forward a saving, and the backup should have been built from the start.