Shoprite's operational update on 12 August 2026 put a number on something South African supply chain managers have been describing anecdotally for two years. Sales through its on-demand platform grew 34.5% to R25.5 billion for the 52 weeks to 28 June 2026. Group merchandise sales from continuing operations grew 7.2% over the same period. The delivery channel grew close to five times faster than the core supermarket business and contributed roughly 36% of the group's total sales growth.

Those are published results from a listed company, and they are useful as market data rather than as a case study. The reason they matter to a manufacturer or an industrial distributor has nothing to do with groceries.

What the number actually measures

A delivery promise of 60 minutes, met consistently, resets what people believe is operationally possible. The person receiving that delivery at home on Saturday is the same person approving your 5 to 7 working day lead time on Monday.

This shows up in three places, and none of them are about speed.

The first is the cut-off. Consumer apps publish an order cut-off and honour it. Most B2B operations publish one and treat it as a target. When a customer's order placed at 14:55 against a 15:00 cut-off ships the next day without anybody telling them, the operation has taught that customer to stop believing the cut-off. The next order goes in at 11:00 with a phone call to confirm, and your customer service team now handles a call that the system should have handled.

The second is promise accuracy. Consumer delivery has trained buyers to expect a date at the point of order, not a range communicated later. An operation that quotes 5 to 7 days and delivers on day 4 has not delighted anybody. It has confirmed that the number it gave was not a real number.

The third is visibility. A tracked consumer delivery costs the retailer very little and removes almost all inbound "where is my order" contact. Most SA B2B operations still absorb that contact as a customer service cost and never attribute it back to the fulfilment process that generated it.

The measurement most operations get wrong

Most South African DCs report on-time-in-full against an internal target date. That number is usually comfortable and usually meaningless.

The useful measurement is on-time-in-full against the date the customer was given at order entry. Those are different numbers in almost every operation, and the gap between them is the size of your credibility problem.

Pull 3 months of order lines. For each, record the promised date given at order capture and the date the customer signed for it. Report the percentage delivered on or before the promised date, and separately, the percentage of orders where the promised date was changed after capture without the customer being told.

That second figure is the one that predicts churn. It is also the one that almost nobody measures.

Where late orders actually go late

An order that misses its promise date usually lost the time inside the warehouse. A stock availability check that was wrong. A pick that could not be completed because the master data said a location held units it did not. A wave release that put the order behind 40 others with no logic beyond capture sequence.

Fleet size and route planning get the attention because they are visible and easy to hold a meeting about. Spending on dispatch when the delay started at order release moves the cost around without changing the outcome.

Trace 20 late orders end to end and timestamp each handoff: capture, credit release, allocation, wave, pick complete, pack, load, depart. In most operations the largest single block of elapsed time sits between capture and wave release, and it is dead time.

Three questions worth asking this month

What percentage of your orders are promised a date the system can actually support? If your ATP check runs on stock on hand rather than stock on hand less allocated less quarantined, the promise is being made on a number that is already wrong.

How many orders per week have their promise date changed silently? Any figure above 5% means your commercial team is managing customer expectations manually, which does not scale and does not survive peak.

What does your cut-off cost you? Moving a published cut-off two hours earlier and honouring it absolutely is usually worth more in customer trust than moving it two hours later and missing it occasionally.

The timing

Black Friday is 27 November. Peak volume does not create service failures. It exposes the ones that already exist at normal volume and are being absorbed by people working late.

An operation that knows its promise accuracy before peak can still decide what to do about it. An operation that finds out in December is managing complaints.

OptiChain Solutions works with South African manufacturers and distributors on order-to-delivery process design, DC operations and service measurement. If you want the promise accuracy number for your own operation before peak, it takes about a week of data work.

Give every order a date you can keep

FarEye gives customers a live delivery date and tracking, and flags late orders before the customer calls. We scope it against your carrier network and current systems.

Book a last-mile review See FarEye

Sources