As of November 15, 2026, the busiest four days in Australian retail are still twelve days away, and the stores that will carry them are already being measured on one number: availability. The Australian Retail Council and Roy Morgan expect shoppers to spend a record A$7.2 billion between November 27 and 30, up 5.5 per cent and A$374 million on last year, with an average planned spend of A$920. Roughly 5.8 million people say they will take part, fewer than the six million last year, but each of them intends to spend more. That combination of fewer baskets and bigger ones is what turns a forecast into an operational problem for anyone running same-day fulfilment.
A peak is a capacity test, not a demand test
Coles gives the clearest read on how much demand now arrives through a screen. Its supermarket ecommerce sales rose 22.9 per cent to A$1.62 billion in the first quarter, and online penetration reached 15.7 per cent from 13.3 per cent a year earlier. Those figures were reported before the promotional window opened, which means the peak will land on a base that is already larger than last year's. The same update noted that shoppers are engaging more heavily with promotions and loyalty programmes, with the value of Flybuys points redeemed up 16 per cent.
The delivery layer has expanded in parallel. Coles runs an exclusive on-demand partnership with Uber Eats offering up to 17,000 products, with plans to extend the service to 500 stores; Woolworths works with DoorDash while continuing its own Milkrun proposition; Aldi and Costco have also partnered with DoorDash. Woolworths has added an A$20 million automated eStore in Western Sydney to raise delivery capacity. Online grocery has already passed discount to become the country's second-largest grocery channel, and penetration is forecast to climb from 10.3 per cent to 17.9 per cent by 2031.
What has not scaled at the same speed is the shelf. A store that promises a thirty-minute window is not promising a warehouse, it is promising that a specific unit sits on a specific facing at the moment a picker walks past. Everything about peak trading makes that promise harder to keep: deeper promotions reset facings, tighter windows shorten the search, and larger baskets mean more lines per pick, each with its own chance of failing.
Where availability actually breaks
The pick face
The pick face is the only place where the digital catalogue and the physical shelf are supposed to agree. When a facings reset happens and nobody re-scans the bay, the app keeps selling a position that no longer holds that product. The gap is invisible in sales data, because the item simply stops selling, and it is obvious in a single walk of the aisle with a phone in hand. During peak, resets happen weekly, so the window between a reset and a re-scan is where most of the missing units live.
The handover
Handover is where a substitution becomes a cancellation. If the rider is already at the door, the store has two options and both cost money: refund the line, or offer an alternative the customer did not choose. Stores that pre-approve substitutions at checkout convert a lost order into a smaller one, which is usually the difference between a complaint and a repeat purchase. The rule is worth setting before the peak rather than during it, when there is no time to explain it to a picker at the door.
| Failure point | How it shows up | Who owns the fix |
|---|---|---|
| Pick face | Order cancelled at the shelf after a reset | Store operations |
| Back room | Item found after the delivery slot has passed | Receiving |
| Handover | Substitution refused at the door | Store and platform jointly |
Reading the three rows together explains why a single availability figure rarely changes anything. A store can improve its pick face score while its back room quietly absorbs the peak, or hold both steady while handover turns recoverable orders into cancellations. The rate only becomes actionable when it is split by cause, because each cause has a different owner and a different fix window.
Report a rate, not a list
Head office does not need a list of missing items, it needs a rate it can compare across stores and across weeks. Three numbers are enough to rank a network and to see whether a fix worked within a month: cancellations per hundred orders, substitutions per hundred orders, and the share of each traced to pick face, back room or handover. Anything more detailed tends to arrive too late to change a shift.
The step that gets skipped is the first one. Definitions drift during a peak, because whoever is on shift improvises, and by December nobody can tell whether a 2.4 per cent cancellation rate is better or worse than the 2.1 per cent recorded in October. Fixing the definitions in writing takes an hour and protects every comparison that follows.
What the price data adds
Quick commerce is not a cheaper channel, and the gap is measurable. CHOICE price comparisons found that products bought through delivery apps cost between 7 and 13 per cent more than their in-store equivalents before delivery fees, and in the most expensive Woolworths comparison the total order came out 39 per cent above the in-store basket. That premium is the reason rapid delivery stays concentrated on urgent top-up missions rather than the weekly shop.
For store teams, the implication is narrower than it looks. If the order is a top-up, the customer came for a specific item and will not accept a similar one, so a substitution is worth less than it is in a planned basket. That argues for treating the two missions differently in the availability data: report the substitution rate for top-up orders separately, and expect it to be lower than the network average rather than higher.
FAQ
Why is availability a store problem rather than a buying problem at peak?
Because in most audits the unit is already in the building, just not where a picker can reach it inside the promise window. Buying failures are solved upstream, while pick face, back room and handover failures are solved inside the store and usually within one shift.
Should a network be judged on cancellations or substitutions?
Both, and separately. A substitution is a recovered order and a cancellation is a lost one, so blending the two hides whether a store is recovering demand or losing it. Report them as two rates per hundred orders and then split each by cause.
How often should the audit run?
Monthly in normal trading and weekly through a peak. The definitions have to stay identical across runs, otherwise the trend is meaningless and a fix cannot be judged against its own baseline.
What is the fastest fix that needs no systems work?
A rule that anything received is either shelf-ready or explicitly flagged as unlisted before the shift ends. It removes the largest single cause of stock that exists in the building but reads as zero to a picker working to a timer.
