Chief Information Officer Japan

Three promises narrowed so that they could be kept

Annual revenue: ¥11.2bn; Year to February 2026; family-owned confectioner
December gift orders: 194,000; Direct-ship oseibo orders in the 2025 season
Selling points: 63; 46 own shops and 17 department-store counters

Decisions & consequences

Moving gift orders off the 1998 system a season before everything else

In the 2021 oseibo season, our merchandising system took 186,000 direct-ship gift orders. Its overnight batch overran on nine December nights, so packing labels arrived late, and 0.38 percent of orders carried a wrong name or address. The system, built in 1998 and maintained by Toyasumi Data Service, also held recipe costing and production plans. One of the two engineers who understood it planned to retire in 2024.

Our managing director for sales and the finance director favoured a single replacement with a cloud package, cut over in March 2023 for ¥820 million, including a ¥70 million discount for taking everything at once. I recommended separating gift orders first. My concern was that a single cutover would meet its first December with merchandising, costing and gift orders all untried at peak.

The board accepted the staged plan in May 2022. We gave up the discount and committed to running two systems for about four years, with interfaces costing ¥58 million and an extended Toyasumi maintenance contract at ¥31 million a year. I asked Toyasumi to stay on as a partner rather than as a vendor being retired.

My team converted the gift-order layouts of 17 department stores into one intake, built a noshi master and re-ran the full 2021 oseibo order file through the new platform twice before going live for ochugen 2023.

The December 2023 peak handled 201,000 orders with no overnight overrun, labels printed the same afternoon, and name or address errors fell to 0.09 percent. The core replacement began in 2025 and is due to go live in March 2027, four years later than the single-cutover plan had proposed.

What changed was the board's expectation that every stage of a replacement must survive a gift season on its own. I would handle the vendor differently. I negotiated documentation of Toyasumi's data definitions only after the extension was signed, and paid ¥6 million for it separately. It belonged inside the extension.

Closing December delivery dates when the carrier's capacity fell

From April 2024, Japan's annual overtime cap for truck drivers took effect. In July, our principal parcel carrier told us that December pickups from our Utsunomiya shipping centre would be limited to 7,500 parcels a day, against a 2023 peak of 10,400, and asked us to reduce date-specified deliveries.

The sales director proposed adding a second carrier across our whole delivery area and keeping every date open. One department-store partner also preferred no change, because its oseibo catalogue was already at the printer. I agreed to a second carrier for Tokyo deliveries only, at ¥14 million, but recommended that our order systems enforce a daily shipping capacity and stop offering specific delivery dates after 12 December. Later orders could choose a delivery window, with ¥200 off shipping.

The president approved this in August 2024. We accepted that some customers would buy elsewhere; sales estimated the loss at about 7,800 orders, or ¥46 million.

My team set the capacity limit inside the gift-order platform, so each channel saw remaining dates in real time, and sent department stores a daily file of closed dates. Shops received a one-page script explaining the window option. The shipping centre planned shifts against the capped volume rather than a hoped-for figure.

We accepted 188,000 oseibo orders in 2024, and 99.4 percent arrived on the date or within the window promised, compared with 97.1 percent in 2023. Overtime at the shipping centre fell by 31 percent. In 2025, under the same rules, 38 percent of online gift buyers chose a window and orders recovered to 194,000.

The board now reviews carrier capacity as a seasonal risk rather than a logistics detail. I would change the timing. Department-store catalogues are printed months ahead, and one partner's catalogue still offered every date, which confused customers at its counters. I should have raised the possibility with partners in spring, before the carrier confirmed its figure.

Reserving production rather than pooling shop stock online

In autumn 2024, our online team proposed showing live shop stock of seasonal sweets so that customers could reserve them for collection at any of our shops. A six-shop pilot ran for eight weeks. Shop stock, however, was counted only at closing, and items sold over the counter before a reservation was picked. In the pilot, 4.2 percent of reservations were cancelled, usually by a telephone call to a customer already on the way.

The online manager wanted to fix this with a midday stock count as well as the closing count. Natsue Mozunuma, our general manager of shop operations, pointed out that many shops open with two staff and that a second count would fall in the lunchtime peak. Some shop managers also liked the extra sales.

I recommended a different design: a fixed allocation for online and collection orders, set at production planning each afternoon, made the next morning and kept apart from shop shelves. Customers would collect from the following day, not the same day. We gave up the immediacy of the pilot and about ¥19 million of forecast sales, and accepted that some reserved boxes would go unclaimed.

My team linked the online store to the production plan, printed reservation labels at the factory and sent the boxes on each shop's morning delivery, with no stock count required. Merchandising set allocations by shop and item, and the handling of unclaimed boxes stayed with shop managers.

Click-and-collect reached all 46 shops in March 2025. Cancellations fell to 0.3 percent. Online and collection sales of seasonal items reached ¥212 million in the year to February 2026, up 24 percent. Unclaimed reserved boxes ran at 1.4 percent, which we report as waste.

The business learnt that a promise can draw on production instead of shelves. I would begin differently next time. I designed first for online demand and stood through a shop's closing routine only after the pilot. Doing that first would have saved eight weeks of cancelled reservations.

Discuss a leadership challenge