Chief Marketing Officer / Japan
Notes from practice: four decisions and their price
- Consumer mobile lines
- 8.9m
- Hinoura Communications, March 2026
- Annual consumer marketing budget
- ¥41bn
- Brand, pricing, insight, measurement and media
- In mobile retail and marketing
- 29 years
- From a shop counter to chief marketing officer
Decisions & consequences
01A cheaper plan under our own name, with its exclusions printed
In autumn 2020 the government's pressure on mobile charges made a lower-priced plan unavoidable for every national operator. The question for Hinoura was how to sell it. Our consumer base then produced ARPU of ¥4,880 a month across 8.6 million lines, much of it from plans designed before data became cheap. A lower price would migrate many customers down.
The finance director and the head of consumer sales proposed a separate low-price brand, sold online. Their argument was sound: it would slow migration from existing plans and protect shop partners, who feared losing traffic. I argued for an online-only plan inside the Hinoura brand. Customers would compare anyway, and a separate brand would let us avoid saying what the lower price removed: free shop assistance, the operator email address, family discounts and bundled options. I proposed that we publish those exclusions on the same page as the price.
The board debated it across two meetings and accepted my proposal with one condition from the finance director that I thought fair: a quarterly report on migration and revenue, with a stated point at which we would revisit the structure. We launched Hinoura Online in March 2021 at ¥3,280 a month for 20GB. The exclusions list had seven items. Shop help, including moving data to a new handset, was charged per visit.
By March 2023 the plan carried 1.42 million lines, 61% of them migrated from our older plans. Consumer ARPU fell to ¥4,190, and we attributed about ¥31 billion of the annual reduction in service revenue to migration. Monthly handset churn fell from 1.08% to 0.81%, and our quarterly number-portability balance moved from a net loss of 38,000 lines to a net gain of 12,000. Complaints about unexpected charges in the first ninety days were 40% lower on the online plan than on shop-sold plans.
The brand now meant the same thing at every price. What I would change: the per-visit charge fell hardest on customers who had moved for price but still needed help moving photographs and contacts. We waived it for data transfer in 2022. I should have tested it with older customers before launch.
Paying the shop counter for explanation instead of attachment
When I became CMO in 2021, our agency partners earned part of their commission for each optional service attached at sign-up: video, cloud storage, device cover and a security application. Many options were free for a month, and customers were told they could cancel later. Many did not. Across our shops 58% of new contracts left with at least one option. Thirty-one per cent of attached options were cancelled within sixty days, and complaints about unwanted options reaching our customer centre and local consumer affairs centres totalled 2,140 in the year.
The stakes were commercial on both sides. About ¥5.7 billion of annual option revenue depended on the practice, and agency companies relied on the attachment payment to fund staff. Our content partnerships team argued that the options were good products customers would not otherwise discover. The two largest agencies said the change would cost them roughly ¥3.1 billion a year. Our head of consumer sales doubted that explanation quality could be measured fairly.
I proposed removing the attachment payment entirely and replacing it with two payments: one for verified explanation, based on a sampled follow-up call within a week of the contract, and one for new customers still with us after six months. To meet the measurement objection, an independent call-centre team ran the sample, and agencies could see and challenge every scored call. Staff could still describe options, but the customer's full monthly total had to be read aloud before signature. We funded ¥2.2 billion of transition support over eighteen months.
Within a year the attachment rate fell to 22%, sixty-day option cancellations to 9% and option complaints to 610. Six-month churn among new customers fell from 7.9% to 5.6%. Option revenue settled about ¥4.9 billion a year lower, and two smaller agencies closed four shops between them sooner than they had planned.
The company's attention moved from what was sold at the counter to what customers understood when they left. I would have designed the scheme with the smallest agencies in the room; their staff costs per shop were higher, and our transition formula under-compensated them.
Switching off television in three prefectures to test the budget
By 2023 our consumer media budget was ¥24 billion, of which 62% went to television. Our attribution model, built with our media agency, Hakuro Media Partners, credited paid search with 44% of customers who joined from another operator and television with 11%. The finance director asked, reasonably, why we did not move television money to search. I distrusted the model, because search was usually the last step of a decision that began elsewhere, but distrust was not evidence.
Our head of marketing science, Ayane Morioka, proposed a withholding test: stop television in three prefectures for eight weeks, compare switching and shop visits with three matched prefectures, and read the results by age. The agency objected that eight weeks was too short to show television's slower effects. Regional sales directors objected that their areas would lose customers to operators still advertising. Both objections were fair. We adjusted the regional sales objectives for the period, ran the test from September to November, away from the spring switching season, and agreed with finance in advance that a television effect below half the modelled figure would move a quarter of the television budget.
The test cost an estimated 6,800 customers in the affected prefectures, roughly ¥330 million of first-year revenue. For customers aged fifty and over, television's incremental effect on switching was about 1.9 times what the model credited. For customers under thirty we could not distinguish television's effect from zero. When television stopped, search volume among older customers fell by about a third, which suggested how much of its credit had been borrowed.
We moved ¥3.6 billion away from national television slots that skewed young, towards regional television and staffed shop events for older customers and online video for younger ones. We now run at least one withholding test a year and report its confidence interval to the board. Neither the agency's view nor finance's first instinct survived intact, and planning conversations became more modest.
I would have run the test for twelve weeks and in more prefectures. One control prefecture saw an unusual competitor promotion, leaving our result for customers aged thirty to forty-nine less certain than I would like.
Asking for consent before building an advertising product
In 2024 our new-business team proposed an advertising product: anonymised audience segments derived from where customers' handsets had been, sold to advertisers for planning and footfall measurement. The forecast was ¥6.2 billion of revenue by the third year. The design relied on notice in our privacy policy and an opt-out setting. Our legal team considered that approach defensible, subject to controls on aggregation and re-identification.
I argued that defensible was the wrong standard. A customer who discovered that their movements had shaped an advertisement would not consult our privacy policy; they would decide whether Hinoura had used them. I proposed explicit opt-in, with a plain description of what advertisers would receive and what they would not, and a monthly 1GB data allowance for participating lines. The head of new business argued that opt-in would yield too little data for advertisers to buy; the finance director noted that the original forecast already sat in our mid-term plan. Both were right, and the board heard their arguments in their words.
The board approved the opt-in design with the forecast cut to ¥1.8 billion by year three. Before launch we tested the consent screen with 600 customers, 150 of them over seventy. The first version explained the mechanism well and the benefit badly; only 13% of test customers agreed. The second described the data allowance first, the advertising use in two short sentences, and exactly where to withdraw.
Twelve months after the April 2025 launch, 21% of eligible consumer lines, about 1.6 million, had opted in, and 4% of those had later withdrawn. First-year revenue was ¥1.3 billion against the reduced plan of ¥1.1 billion. In our annual trust survey, the share of customers comfortable with how Hinoura used their data rose from 38% to 47%, although other changes that year also contributed.
New data products at Hinoura now begin from consent rather than notice, and the privacy office reviews consent screens before any business case is approved. I would have tested the first screen with older customers earlier still; they read it more carefully and found the gaps in our explanation faster than our designers did.