Renting the system, still owning the promise

At 07:40 on a Tuesday in November 2024, Kuranoe Information Systems told its 23 user firms that the overnight batch had stopped partway. Positions and cash balances still reflected the close of business on the previous Friday; Monday's trades and payments had not been posted. Kuranoe expected recovery at about eleven, long after our branches opened at nine.

Three options were on the table by eight. Tetsuo Hagimura, our head of retail sales, wanted to accept every order and have branch staff check Monday's trade confirmations by hand. Our compliance general manager preferred to suspend all orders until recovery, so that every customer was treated alike. I proposed a third course: accept orders from accounts with no trades or cash movements on Monday, whose Friday positions were still accurate, and hold the 1,180 accounts that had changed. We could identify them only because we keep our own copy of each day's executed orders. The president approved my proposal at 08:15. Its cost was that the most active customers, who might most want to act again, waited longest.

Branch managers received the list of held accounts by 08:40, with a script explaining why an order could not yet be taken and when we expected to call back. Kuranoe restored positions at 11:05, and we released the held accounts at 11:20 after reconciling a sample against our own records.

No settlement failed, and no order was accepted against shares a customer no longer held. We received 214 complaints, mostly from held customers, and two took their case to the industry's dispute-resolution service. We made no goodwill payments for price movements during the hold. Afterwards Kuranoe agreed to deliver a position file by 06:30 every morning whatever the state of the batch, and the rule we improvised is now written policy. Next time I would want that rule agreed with sales and compliance before any incident, and I would tell held customers by text message before nine rather than at the counter.

In the spring of 2025, criminals used phishing sites to take over customer accounts at several Japanese online brokerages and place unauthorised trades. Our online channel had 41,000 enabled accounts, protected by a password and a one-time code sent by email or SMS. We had suffered no takeover, but our customers received the same phishing messages, and a one-time code can be typed into a fake site as easily as a password.

In May I recommended that from October every online sign-in require a passkey on the customer's phone or a hardware security key, with no SMS fallback. Daichi Morosawa, who runs our online services, argued for keeping SMS codes for customers whose phones could not hold a passkey; he expected several thousand to leave the channel. I accepted that the channel would shrink. A fallback that phishing could defeat would simply have become the route attackers used, and the president agreed.

We spent ¥118 million on development, 6,000 hardware keys and extra contact-centre staff. Advisers helped customers enrol during home visits and at branches through the summer. Customers who could not or would not enrol were offered telephone ordering at online commission rates for twelve months, which we estimated would forgo about ¥46 million of commission.

By the October deadline, 33,800 customers had enrolled passkeys and 2,100 were using hardware keys. Of the remaining 5,100, 3,400 chose telephone ordering and 1,700 did not respond; their online access was suspended until they enrolled at a branch. Contact-centre calls rose 31 per cent for nine weeks. We have had no online account takeover since, though I cannot claim the change alone explains that. I would have tested enrolment with customers over seventy-five before announcing the date. I would also have presented hardware keys as an equal option from the start, rather than as a fallback that some customers felt marked them out as less capable.

In the autumn of 2022, the board had ¥640 million of change budget for the following year and two serious claims on it. Our planning office wanted a customer app ready for the new NISA in January 2024, aimed at younger investors who were opening accounts with online brokers. I proposed an inheritance case system first. We were handling about 2,900 inheritance transfers a year, with a median of 94 days from first contact to transfer. Heirs, many living in Osaka or Tokyo, received repeated requests for family-register documents, and only 38 per cent of the value we transferred stayed with us.

The stakes ran both ways. The app was part of how the firm would reach the next generation of customers; the inheritance work concerned assets that were already ours to lose and families at a difficult moment. Our head of planning disagreed strongly, and the president asked us to put both cases to the board. The board funded the inheritance system at ¥410 million and deferred the app by fourteen months.

We built a case system that listed the documents required for each family circumstance, checked family-register documents centrally rather than in branches, let heirs follow their case status by telephone or online, and sent transfer instructions to Kuranoe without rekeying. The estates desk in Takamatsu took over cases from branches, which some branch managers resented, because the heirs had been their relationships.

By 2025 the median transfer time had fallen to 41 days, and 52 per cent of transferred value stayed with Ogiwatari. The cost was real: the app arrived in March 2025, and we opened 9,800 new NISA accounts in 2024 against a plan of 16,000. I cannot know how many more the app would have brought. I would change two things. I underestimated how many heirs lived outside Shikoku and added online document submission late. I should also have offered a simple NISA application form in 2023 instead of deferring every digital feature to the full app.