Ecommerce inventory management
Our products › Ecommerce inventory management
Small orders, high line counts, a returns rate that would frighten a wholesaler, and a cut-off that never moves.
Everything below assumes the stock figure already carries location, condition and commitment rather than a single total — that argument sits on the pillar. Inventory management software.
The e-commerce order profile
One or two lines per order, arriving constantly, in a mix nobody can forecast at item level. That profile breaks methods designed for pallet movement, and it breaks them quietly.
A wholesale operation moves few orders with many lines each, so the walk is amortised across a large pick. An e-commerce operation moves many orders with one or two lines, so the walk is the job. Route, slotting and batching stop being refinements and become the whole cost base.
In e-commerce, the distance between picks is the cost. Everything else is rounding.
Returns are inventory
Twenty to forty percent in some categories, and every returned unit is stock you own, cannot sell, and cannot see until somebody has looked at it.
The useful measure is the time from arrival at the door to sellable on the channel. In most operations nobody measures it, and in most operations it is days — which means a measurable part of the range is invisible at any moment.
It needs disposition states rather than a single returns bin: arrived, inspected, back to sellable, to repair, to liquidation, written off. Each is a different figure and only one of them may be promised to a customer.
A return sitting in a tray is not an operations problem yet. It becomes one when you buy goods you already own because the system cannot see it.
The window
Most of the day's orders land in the last two hours before cut-off. Capacity has to be sized against that window, not against the daily total — a comfortable day's volume is an impossible two hours.
Which is why the honest question about a pick method is not how many lines an hour it does on average, but what it does in the ninety minutes when everything arrives at once and one picker called in sick.
Channel updates and the last unit
The one place this page overlaps with allocation, and it is handled there rather than repeated here.
Briefly: the interval between a pick confirmation and the channel's availability changing is a window in which the last unit can be sold twice, and it is a timing question rather than an accuracy one.
Peak
November doubles everything, including the number of people in the building who do not know where anything is. Temporary staff are not the problem; expecting them to carry warehouse knowledge is.
A directed operation degrades gracefully here because the system carries the knowledge. A new picker is told the location, scans it, and is stopped when the scan disagrees. Onboarding drops from a week of shadowing to an hour, and the error rate of a temp stops being materially different from the error rate of a regular.
What breaks first as you grow
Usually not the software, which is the uncomfortable part of the answer.
The item master goes first — barcodes that were never right, variants that share a code, dimensions nobody filled in. Then the pick face, because a range that doubled is still allocated the same frontage. Then the process area, when packing, returns and dispatch compete for the same floor at the same hour.
Each of those is fixable, and each takes longer than installing software. The sequence is worth knowing before you conclude that the system is the constraint.
If growth hurts, check the item master before you check the software. It is the cheapest thing to fix and the commonest thing to be wrong.
Frequently asked questions
What is ecommerce inventory management?
Managing stock for an order profile of one or two lines at high frequency, with a returns rate in the tens of percent and a fixed daily cut-off. The demands differ from wholesale mostly in pick method, returns handling and how capacity is sized against the last two hours of the day.
How should returns be handled as inventory?
As owned goods in a non-sellable condition, with explicit disposition states between arrival and being sellable again. The measure worth tracking is the time from the door to available on the channel; in most operations it is days, and in most operations nobody knows the figure.
Why does our warehouse struggle only in the last two hours?
Because that is when most of the day's orders arrive. Capacity sized against the daily total looks adequate and fails against the window. The useful question about any pick method is what it does at peak with a picker missing, not what it averages.
What breaks first when an e-commerce operation grows?
The item master, then the pick face, then the process area — usually in that order, and usually before the software. Each takes longer to fix than a software change, which is why the sequence is worth knowing before concluding the system is the constraint.
Your order profile, honestly assessed
Half an hour on lines per order, returns rate and your cut-off — and a straight answer on whether it is a software problem yet.