A household's viewing record should be used only for purposes that household could read, understand and refuse.
01 / Chief Data Officer
Annual company revenue
¥23.8bn
Year to March 2026; cable, broadband, community channel and resold mobile
Subscribing households
231,000
Of 520,000 homes passed in Nara and southern Kyoto prefectures
Data colleagues
34
Engineering, analytics and the privacy and data-governance office
＋
Designed for consequence.
Cable subscribers rarely think of their set-top box, their router or a call to our help desk as data, and they should not have to. My work is to decide which uses those records can bear, to measure honestly what we learn from them, and to keep any use a household would find surprising out of our plans until that household has agreed to it.
I treat each viewing log, churn score and call summary as a promise to a household, and approve a use only when we could explain it to that household at its own front door.
I am chief data officer and an executive officer of Tamakazura Cable Networks, which provides cable television, fibre and coaxial broadband, a community channel and resold mobile service across Nara Prefecture and the hill towns of southern Kyoto Prefecture. We pass 520,000 homes and serve 231,000 subscribing households. Revenue for the year to March 2026 was ¥23.8 billion, with operating profit of ¥2.6 billion and 940 employees. I lead 34 people in data engineering, analytics and a small privacy and data-governance office, and I report to the president.
Most of my career has been spent measuring audiences. I began in 1999 at a survey company in Tokyo, weighting diary and online panels, and later ran audience research and then catch-up streaming analytics for an Osaka television broadcaster. Those years taught me how easily a count can be made to look larger or more certain than it is. Since 2016 I have worked at Tamakazura, where the people we measure are also customers who pay us every month and can telephone the same afternoon if something we do with their data surprises them.
The decisions on this site involve money we chose not to take, or took more slowly. I insisted on opt-in consent before viewing logs went to an advertising partner, which reduced a ¥140 million annual offer to about ¥39 million. I retired the strongest signal in a churn model I had built myself, because it was paying discounts to households whose real problem was a faulty line. I also held an AI call-summary tool to one contact centre until we could show where it made mistakes. Each choice had an internal opponent with a fair argument.
28.5% of eligible households chose to share viewing logs
After nine months of on-screen and postal requests, 33,600 of 118,000 capable set-top boxes had opted in to partner measurement. The smaller pool cost revenue, but every outside use since then rests on a consent record we can show the subscriber concerned.
Repairs came before discounts for repeat-fault households
Splitting churn-model output into a repair route and a price route cut monthly churn in the repair group from 2.7% to 1.5% over a year, while retention discount spending fell from a planned ¥260 million to ¥158 million.
Call summaries were audited before anyone trusted them
A monthly audit of 400 AI-drafted summaries found material errors in 3.4%. Complaint classification therefore stayed with agents, while after-call work in the pilot centre fell from 6.1 to 4.0 minutes.
Current focus
I am replacing our partner data agreements, which grew one at a time, with a single register that records each partner's purpose, the minimum fields shared, the retention period and the date on which someone must decide again whether the arrangement should continue.
Conversations I welcome
I would consider one outside directorship in a Japanese consumer-service company that holds household data, offering about twenty-four days a year once my president has approved the commitment and the relationship checks are complete.
The leadership record
Trusted with consequential decisions.
Formation
Bachelor of Economics (statistics seminar)Hagimine University · 1999
Haruka brought the viewing-log proposal to me with the revenue we would lose already calculated. She did not make the decision easy, but she made it clear, and she has never hidden a mistake in her own work from me.
Masahiro UneshimaPresident, Tamakazura Cable NetworksThe full record ↗
Contact
I would consider one outside directorship in a Japanese consumer-service company that holds household data, offering about twenty-four days a year once my president has approved the commitment and the relationship checks are complete.
