Decisions that asked customers to choose, and what we paid for them

Declining a redelivery fee and changing the default instead

When I joined Kazemachi in April 2021, 13.6 per cent of our parcels needed a second attempt: about 55 million extra doorstep visits a year. Operations estimated that without a change we would need around 2,300 more drivers once the overtime cap for drivers took effect in April 2024.

Our executive officer for operations proposed a fee of ¥110 for every second or later attempt, charged to the recipient. His model showed ¥4.1 billion a year of revenue and a fall of perhaps a third in missed deliveries through deterrence. The chief financial officer supported it. I argued against. Most recipients had not chosen the carrier, the day or the window; the online shop had, often by default. A fee collected at the door would put the anger on our drivers and the cost on the wrong person. I proposed instead a ¥5.2 billion programme over three years: evening-before notices for every reachable recipient, one-tap choices, 4,500 additional lockers and a credit to shippers whose recipients chose a pickup point at checkout.

The president asked for a trial with a condition I accepted: if the second-attempt rate was not below 10.5 per cent by the year to March 2023, the fee would return to the board. We also accepted a cost the fee would have avoided. Unattended placement at the door, which many customers chose, brings claims for missing parcels, now about ¥310 million a year.

The rate fell to 10.1 per cent in the year to March 2023 and to 7.9 per cent by the year to March 2026. Lockers grew from 1,050 to 5,600, and pickup points took 14.1 per cent of parcels, up from 2.8 per cent. We estimate the saving at about 2.1 million driver hours a year. The fee was never introduced. Each depot's second-attempt rate is now reported monthly, and depot managers and my team answer for it together.

I would do two things differently. The programme assumed a smartphone, and in our rural depots the rate, which began lower at 9.8 per cent, fell only to 8.9 per cent; the telephone and paper routes should have been built first. I would also have shared the fee model openly with the board's outside directors rather than presenting only my alternative.

Publishing the reasons for the April 2024 price revision

By the autumn of 2023 the cost of carrying a parcel had moved well ahead of its price. The overtime cap from April 2024 meant we needed about 1,900 more drivers to cover the same routes, base pay for drivers was rising 7.5 per cent, and subcontracted trunk haulage and fuel had both become more expensive. Finance calculated that revenue per parcel had to rise by about 10 per cent. Seventy-one per cent of our volume came from contract shippers; the largest, an online marketplace called Kurumi Bazaar, sent us 19 million parcels a year.

I proposed that we announce the revision five months ahead and publish the reasons: an explanation, in plain words and approximate shares, of what had changed in our costs. I also proposed a structure that rewarded cheaper behaviour: a ¥60 discount for individual senders who addressed a parcel to a locker, and contract increases that varied from 6 to 18 per cent with each shipper's share of pickup-point deliveries and the accuracy of its volume forecasts. Our executive officer for corporate sales disagreed strongly. Publishing reasons, he said, would invite every shipper to dispute our numbers, and negotiation should stay private. The chief financial officer wanted a uniform increase without the locker discount, which would cost about ¥1.1 billion a year. The president backed publication and the structure but reduced the discount from ¥80 to ¥60.

Over-the-counter rates rose by an average of 9 per cent in April 2024 and contract rates by 12 per cent. Kurumi Bazaar declined anything above 8 per cent and moved all its volume to two competitors from July 2024. Our volume fell 3.5 per cent that year. Seventeen of our twenty largest shippers renewed. Revenue per parcel rose from ¥668 in the year to March 2024 to ¥733 in the year to March 2026, and operating margin from 2.9 to 4.6 per cent. We now send every contract shipper an annual letter explaining how our costs have moved.

I would change two things. I wrote the explanation for consumers first and shippers second; it should have been the reverse. And I would have met Kurumi Bazaar's chief executive myself in the summer, before its buyers read our letter.

Taking the links out of our own text messages

From 2022, criminals sent waves of text messages in Kazemachi's name, claiming a parcel could not be delivered and linking to sites that stole card details or installed malicious software. By June 2023 our contact centres were taking 41,000 calls a month about suspect messages. Our own research found that 38 per cent of recipients now ignored genuine Kazemachi texts. That mattered commercially, because our redelivery programme ran largely on those texts: 31 per cent of all delivery-choice changes came through a link in one of them.

I proposed in September 2023 that Kazemachi stop putting any web link in its text messages. Our texts would carry the tracking number and an instruction to open the app or call the number printed on the slip, and we would tell the public, repeatedly, that Kazemachi never sends links by text. The head of digital channels in my own team disagreed, as did the executive officer for operations. Both expected missed deliveries to rise, and they were right.

We made the change in October 2023. The second-attempt rate rose from 9.3 per cent to 10.2 per cent for two quarters, roughly 3.5 million extra visits. We launched an automated telephone line for delivery choices in January 2024, four months later than it was needed, and pushed app registration through depots, slips and our shippers' order confirmations. Registered app users rose from 5.2 million to 9.4 million by March 2026. Calls about suspect messages had fallen 72 per cent by December 2024, and the share of recipients who said they ignored genuine notices fell from 38 to 17 per cent. The second-attempt rate was back below 9 per cent by the end of 2024.

What changed was our rule for customer messages. Any new notice is reviewed against the question of whether a criminal could copy it convincingly, and the channel rule is written policy reviewed by the board's risk discussion each year.

I would have acted in 2022, when the first wave appeared, rather than waiting for the call volumes to prove the point. And I would have launched the telephone line on the same day we removed the links, so that customers without the app were never left without a route.

Refunding cancelled fares in cash when credit would have protected the quarter

In April 2020 Sorahane Airways faced the collapse of domestic travel. Bookings for May were down by nearly 90 per cent, and ¥7.6 billion of tickets already sold for flights between April and August were being cancelled, by passengers or by us. The airline had about ¥14 billion of cash and fixed costs of about ¥6 billion a month. Every refund was cash the company might need to survive.

Our chief financial officer proposed that refunds be issued as flight credit valid for twelve months, with cash available only on written request. I was general manager of marketing and loyalty, and I argued the opposite in a paper to the board: cash to the original payment method by default and without fees, an optional credit worth 110 per cent of the fare for those who preferred it, and a one-year freeze on members' status. My case was that our customers were mostly Kyushu residents who would fly with us again only if they trusted us now, and that a credit imposed by default would be remembered long after the cash crisis passed.

The board approved the cash default and the 110 per cent credit but set the credit's validity at eighteen months, not the twenty-four I had proposed, and allowed up to sixty days to process cash refunds. Sixty-four per cent of affected customers took cash, and refunds paid out reached ¥4.9 billion. The airline drew on a committed bank facility of ¥20 billion earlier than it had planned, and the finance team bore the consequences of my recommendation in every cash forecast that year.

Complaints about refunds stayed low, and our consumer-affairs log recorded fewer than 150 formal complaints across the period. By March 2021, 58 per cent of members who had been active in 2019 had booked at least one flight again, and the credit was redeemed at a higher rate than the finance team had modelled. The cash default was written into our conditions of carriage as standard practice.

I would change two things. Eighteen months proved too short once the second and third waves arrived, and we had to extend the credit in January 2021, which looked like a concession rather than a policy. And sixty days created a backlog that a larger refund desk, staffed in March rather than May, would have avoided.