An empty tray at eight o'clock is a sign of good planning

At ten o'clock on most nights, in most bakery chains in Singapore and Malaysia, someone in an apron slides the remaining buns off the trays and into bags. Some go to staff. Some go to a food-rescue charity if one collects nearby. The rest go into the bin. The next morning the same outlet bakes the same quantities, because the plan says it should, and because a full display at closing time has long been treated as a sign of a well-run shop.

I have come to think the opposite. A display that is still full at closing time is evidence that the morning plan was wrong, and that the outlet will repeat the error tomorrow. A few empty trays in the last hours, on the right lines, are evidence that someone planned well.

The full-shelf rule and its cost

Rumbia Bakehouse outlets bake on site from frozen dough and part-baked products made in our Senai and Senoko kitchens. For years each outlet baked three times a day, at about six in the morning, eleven and four in the afternoon. The rule handed down to managers was simple: no empty tray positions before closing. Our mystery-shopping programme marked outlets down for gaps, and managers' bonuses reflected those scores. Nobody's bonus reflected what was thrown away.

The result was predictable. Managers baked the four o'clock batch large enough to fill the display until ten at night, knowing that much of it would not sell. Across the network, unsold product reached 9.4 per cent of production value in 2024. Some of it reached people who needed it through Lantern Table, the food-rescue charity whose board I sit on, and I am glad of that. But a charity collection is not a reason to overproduce, and much of what we threw away never reached a collection at all.

Waste was not the only cost. A large afternoon bake heats a small outlet at the hottest time of day, when our display chillers are already struggling. It occupies staff during the after-school trade. And it produces bread that will sit for five or six hours before it sells, which is not the bread we want customers to remember.

Baking in smaller waves

In 2025 we began a trial in 40 outlets. Instead of three large bakes, those outlets bake five smaller ones, each sized from the previous fortnight's sales for that hour on that day of the week. The final bake is the smallest. After eight o'clock, managers may let up to a third of tray positions empty for slower lines, provided that a short list of core items, our white loaf, two buns that families buy for breakfast and the cakes ordered in advance, remains available until closing.

The first weeks were awkward. More bakes meant more oven starts and more interruptions for a small team, and two outlets ran out of their most popular buns before six in the evening because the model had not yet learnt their pattern. Some managers were openly uncomfortable leaving gaps on the display, having been rewarded for years for the opposite. Our regional managers had to tell them, repeatedly, that an empty tray at eight o'clock would not cost them anything at review.

Smaller bakes also helped with heat. An outlet that runs its oven for one large batch at four o'clock adds heat to a small room at the hottest point of the afternoon, just when the display chillers are working hardest. Spread across five bakes, the oven load at any one time is lower, and in two kiosks in transport stations our loggers showed the air behind the counter running about two degrees cooler in the late afternoon. The staff noticed before the loggers did.

By the end of the year the trial outlets had cut unsold product from 9.4 to 6.1 per cent of production value. Sales in their last two trading hours fell by about 1.8 per cent against similar outlets that kept the old rule, but their total sales were within half a percentage point of those outlets, because fresher afternoon bakes sold better earlier in the evening. Lantern Table now collects less from those outlets. I declared that to its board, and we agreed it was the right result for both organisations.

Counting the last hour honestly

The deeper change was in measurement. A rule about full shelves measures the display. It does not measure the customer. We replaced the empty-tray score with three measures that describe what customers and the business actually experience.

The first is the sell-out time for each product line in each outlet, recorded automatically when the last unit is scanned. A core item that sells out before closing is a failure we should fix. A slow line that sells out at nine o'clock is a success.

The second is unsold product as a share of production value, by outlet and by line, reported every week to the outlet manager and every month to me. It is now part of every manager's bonus, beside sales and food-safety scores.

The third is what customers say. We added a question to our receipt survey about whether the customer found what they came for. Scores in the trial outlets have not fallen, which suggests that most customers in the last hour want a loaf or a few buns for breakfast, not the full range.

None of this is new to people who run bakeries well. Independent bakers have always baked to the day and accepted that they will sell out of some things. Chains forgot it because a uniform display is easier to inspect from head office than a thoughtful plan. The full-shelf rule is a rule for auditors.

There are limits. Some formats depend on abundance: a bakery in a transport hub at the evening rush may lose more from gaps than it saves in waste, and our model treats those outlets differently. A trial of 40 outlets over one year is not proof, and I want to see the results hold through the Lunar New Year and Hari Raya seasons before we change every rota.

For directors of any food retailer, the question is simple to ask. Does the company measure what it throws away as closely as it measures what it sells, and does anyone's pay depend on the first number? If not, the shelves will stay full and the bins will stay full with them.