Chief Strategy Officer / Japan
Three decisions and what each one cost
- Net sales
- ¥142.6bn
- Year ended February 2026; listed regional supermarket chain
- Supermarkets
- 71
- Across Iwate, Akita and Aomori prefectures
- Mobile sales vans
- 24
- Serving towns and districts without a nearby store
Decisions & consequences
01Closing two inland stores and paying for van routes in their place
In 2022 two of our stores in inland Iwate towns faced the same decision. Both towns had lost about a quarter of their residents since 2005, and their combined sales had fallen to ¥1.31 billion with an operating loss of ¥96 million. Both needed refrigeration replacement within eighteen months, at a combined cost of ¥340 million. A drugstore chain had opened in each town, selling much of the dry grocery our stores relied on.
My office set out three options: renew both stores, close both, or close both and serve the towns through mobile sales vans and a staffed pick-up counter. The store-operations director favoured outright closure. He argued that vans lose money and that customers would simply drive to our larger store twenty-five minutes away. One outside director was concerned about how closure would be received locally. I recommended the third option because about a third of the stores' regular customers were over seventy-five and many no longer drove.
The board approved closure from March 2023, nine new vans costing ¥54 million and a pick-up counter in one town's community centre two days a week. We gave up about ¥500 million of sales that the drugstores and other retailers absorbed. Of 38 part-time colleagues, 29 transferred to other stores or to van driving and preparation; nine left the company.
In the year to February 2026 the nine routes sold ¥620 million and lost ¥41 million, compared with the ¥96 million store loss and the renewal we avoided. About 1,900 customers buy from the vans at least twice a month. The pick-up counter has been less successful, with fewer than forty orders a week, and we are reviewing it.
What changed was the company's willingness to treat closure as a decision about service rather than retreat. If I did this again, I would agree the order-taking arrangement for housebound residents with the town's welfare council before the closure date, rather than three months afterwards.
Retiring most of our own label to join a shared private brand
By 2022 our own label carried about 900 items and 6.1 per cent of net sales. It was too small to give manufacturers volume, and larger chains from outside the region were selling private-brand staples below our cost. Two neighbouring chains, Akanemori Stores in Akita and Utsumineya in Aomori, faced the same problem. Together we bought about ¥38 billion of the categories we could share.
I recommended that we form a joint purchasing company with them and develop one shared private brand. Our merchandising director, who had built the own label, opposed retiring it; he believed customers associated it with our stores. Several local makers worried that joint buyers would delist them. Members of the founding family asked whether we were taking the first step towards a merger they had not agreed to.
We created Mitsurane Procurement in April 2023, each chain holding one-third of its ¥300 million capital. We retired 640 own-label items over eighteen months and kept about 260, mostly fresh and regional products. Each chain retained a protected local shelf of around 180 products that the joint buyers could not remove. The shareholders' agreement stated that the company would not become a vehicle for a merger without a separate decision by each board.
By the year to February 2026 the Tsumugiba range carried 1,150 items and 8.4 per cent of our net sales, with gross margin 3.1 points above the branded products it replaced. We lost the business of two small suppliers whose products overlapped with shared items. Our merchandising director moved to lead the protected local range, which has grown faster than the rest of grocery.
The arrangement changed how we think about scale: we can buy like a larger chain without becoming one. What I would do differently is write the rules for product-specification disputes before the first product launched. Three disagreements in the first year, one over the salt content of pickles, took months because nobody had the final say.
Choosing staff-assisted checkouts over full self-checkout
In 2024 our stores were about 310 full-time equivalents short of the part-time hours they needed, and checkouts consumed 34 per cent of store labour. Older checkouts in many stores were due for replacement. The question was what to replace them with.
Our finance director proposed full self-checkout in every store, at an investment of ¥2.6 billion, saving an estimated 410,000 labour hours a year. I proposed a mix: staff-assisted checkouts, where a colleague scans and the customer pays at a machine, in most stores, and full self-checkout only in the 14 stores with the youngest customer base. Our point-card data showed that customers aged sixty-five or over provided 46 per cent of sales, and store managers reported that many of them asked for help even with payment machines. The mixed plan cost ¥1.7 billion but saved only about 290,000 hours.
The board accepted the mixed plan. We gave up about 120,000 hours of annual savings, worth roughly ¥130 million at our average cost per hour. In exchange we kept a person at every checkout for customers who wanted one, and we limited the theft exposure that full self-checkout brings in stores without the staff to supervise it.
By February 2026 the new checkouts were installed in 63 stores, and the actual saving across them was about 305,000 hours, slightly above the estimate. We moved 180,000 of those hours into prepared-food production in stores and at the process centre, and prepared-food sales rose 11 per cent over two years. Shrinkage in the self-checkout stores rose by 0.2 points of sales, as finance had warned.
The decision taught us to allocate hours by what customers notice, not only by what machines can do. If I repeated it, I would run a two-store trial first and measure how older customers actually used each design, instead of relying on managers' reports to set the split.