Is Takealot's Replenishment Algorithm Costing It Millions in Lost Revenue?

When a product sells out, a lack of subsequent sales does not necessarily mean a lack of demand. Yet Takealot's replenishment system appears capable of treating it that way.

A system preventing supply

Takealot will not allow me to send it stock of a product because, according to its system, there is "insufficient demand."

There is just one problem: the product has not been selling because Takealot had no stock available to sell.

This may sound like an isolated seller frustration, but variations of the same complaint appear repeatedly in South African e-commerce groups. Sellers report products being blocked from replenishment after selling out, leaving them unable to restock precisely because there were no sales during the period in which the products were unavailable.

If that behaviour is occurring at scale, it raises a much bigger question: how much revenue is Takealot losing because its own replenishment system is preventing sellers from supplying products customers may still want to buy?

A real example

Earlier this year, I launched a new product on Takealot and sold 22 units in June.

By mid-July, our remaining stock at Takealot had sold out. We were waiting for a larger shipment that was expected to arrive in early August, but delays at South Africa's ports meant the stock only arrived towards the end of the month.

When the shipment finally arrived, I attempted to replenish the product. Takealot's system blocked the shipment, citing "insufficient demand."

But how can demand be accurately measured when a product has been out of stock?

Surely the most relevant period is the one in which the product was actually available: from the date it was last replenished until the final unit was sold. The weeks after it sold out cannot provide a meaningful measure of sales demand because customers had nothing to purchase.

The same problem applies if page views form part of the calculation. An out-of-stock product will naturally attract fewer views. It may become less visible in search results, and customers have little reason to open a listing for an item they cannot buy.

Low sales and low page views during an out-of-stock period are therefore not necessarily evidence of insufficient demand. They may simply be evidence of insufficient stock.

The algorithm appears to be missing context

Takealot introduced replenishment limits for an understandable reason. Before Amazon entered the South African market, Takealot faced the difficult task of managing fulfilment centres that were receiving more seller stock than they could efficiently accommodate.

A system that limits slow-moving or excessive inventory is sensible. Warehouse capacity is finite, and allowing sellers to send unlimited quantities would create its own operational problems.

The concern is not that replenishment limits exist. It is whether the algorithm is distinguishing between a product that is genuinely not selling and one that sold out and was never replenished.

A more accurate calculation should consider questions such as:

• Did the seller's previous stock sell out?
• How quickly did it sell while it was available?
• When was the product last replenished?
• Was any stock available during the period used to calculate demand?
• Did sales and page views decline only after the product became unavailable?
• Was the product available on lead time during the out-of-stock period?

Without this context, an algorithm can create a self-reinforcing loop: a product sells out, its sales and page views fall, the system concludes that demand is insufficient, and the seller is then prevented from sending the stock needed to generate new sales.

The workarounds create further problems

Sellers generally have two possible workarounds.

The first is to log a support ticket and request that the product be reopened for replenishment. When approved, the seller may receive a limited window in which to create and send a shipment before the system blocks the item again.

However, these requests are not always approved. More recently, I have had requests rejected because support staff appeared to accept the system's decision as correct without properly examining the product's stock history or considering whether the algorithm itself could be wrong.

The second option is to place the product on lead time. When an order arrives, the seller sends an individual unit to Takealot for onward delivery to the customer.

This may keep the listing available, but sending products one unit at a time significantly reduces seller margins. It is a workaround, not an efficient or commercially sustainable fulfilment model.

There is also a strange consequence: once a single lead-time order arrives, Takealot's system may suddenly recognise that demand exists and allow the seller to replenish the product again.

In other words, the seller may need to accept a low-margin order to prove demand for a product whose earlier stock had already sold out.

The Amazon contrast

The commercial impact became particularly clear when my delayed shipment arrived.

As soon as the new stock was captured in our inventory system, we began receiving Amazon FBM orders. Those orders could be fulfilled directly and profitably from our own warehouse.

The contrast was striking.

Amazon saw available stock and allowed customers to buy it. Takealot saw no recent sales, concluded that there was insufficient demand and prevented us from sending the same product to its fulfilment centres—even though the lack of recent sales was caused by Takealot having no stock.

This does not prove that every replenishment decision is wrong, nor does one product establish the total financial impact. However, when many sellers describe the same pattern, it deserves proper investigation.

A potentially expensive blind spot

Sometimes the problem is not a lack of customer demand. It is a system standing between the customer and the product.

Every blocked replenishment potentially means fewer products available, fewer customer orders, less seller revenue and less GMV for Takealot. Across thousands of sellers and listings, even a relatively small error rate could translate into a substantial commercial loss.

Could the cumulative impact already amount to millions of rands in lost revenue? Without access to Takealot's internal data, it is impossible to know. But the possibility is credible enough to warrant attention.

Takealot should review how its replenishment algorithm treats out-of-stock periods and ensure that support teams can meaningfully reassess questionable decisions rather than simply deferring to the system.

Algorithms are useful for managing scale, but they should not be treated as infallible—particularly when their decisions may be preventing willing customers from buying products that sellers have ready to supply.