Retail & E-commerce Logistics - Speed, Accuracy, and Returns Mastered

One inventory pool, several channels, conflicting promises

Selling through marketplaces, a direct-to-consumer site and a B2B channel from one stock pool is the right design commercially and it creates a conflict the warehouse has to resolve every day: the channels have different service promises, and during a shortage they cannot all be honoured.

Marketplace dispatch SLAs carry account-level penalties and account health consequences. D2C orders carry the brand relationship directly. B2B replenishment carries the largest individual order values. Left undefined, the resolution is made by whoever is picking, which means it is made differently each time and cannot be explained to anyone afterwards.

The answer is an allocation rule agreed commercially and implemented in the WMS rather than in someone's judgement - which channel is protected when stock is short, and at what threshold the rule changes. It is a fifteen-minute decision that prevents a recurring argument, and almost nobody makes it until after the first bad week.

• Write the allocation rule before peak, not during it.
• Implement it in the system, not in picking-floor convention.
• Review it after each peak - the right answer changes as channel mix does.

Peak is a shape, not a number

Festive volumes running four to eight times normal are usually planned against a single multiplier, which is the least useful way to describe them. Two operations with the same peak multiple can need completely different designs.

What determines the design is the shape: how long the elevated period lasts, how sharply it starts, how much notice there is, and - the part most often left out - when the returns wave arrives. A promotion generates its returns two to four weeks later, by which time the operation has usually stood down its temporary labour. The outbound peak is planned for; the reverse peak that follows it frequently is not, and it lands on a team that has just been reduced.

Planning both waves together is the single change that most improves a peak season. It affects when temporary staff are released, how much grading capacity is held back, and whether the returns process is still standing in the week that decides how much of the season's revenue is actually kept.

What marketplace SLA compliance actually measures

Marketplace performance metrics do not measure whether an order was handled well. They measure whether specific events were recorded within specific windows, and the difference matters because it is possible to fail while doing everything right operationally.

The usual causes of a compliance failure are not picking or packing. They are a scan that happened but was recorded late, a manifest closed after a cut-off, a handover to a courier that occurred before the system reflected it. These are integration and process-timing faults, and they are invisible to anyone measuring warehouse throughput.

So a fulfilment operation selling through marketplaces needs to monitor the event timestamps the marketplace reads, not only its own dispatch performance. Where the two diverge, the marketplace's view is the one with commercial consequences, and it is the one that determines whether the account stays healthy.