The shape of a pick, in two parts
Every manual pick order splits into work that scales with the number of lines and work that scales with how spread out those lines are. Reaching, counting, scanning and placing is the first kind: twelve lines take twelve picks whether they sit in one aisle or twelve. Walking is the second kind, and it is almost entirely a function of geography.
Two pieces make up the walking. Inside an aisle, a picker goes as deep as the deepest pick and comes back, so the distance is twice the depth of the furthest one. With picks landing at random along the aisle, the expected depth of the deepest of k picks is k over k plus one of the aisle length — one pick averages half way in, two picks two thirds, three picks three quarters. Across the building, the picker has to reach the furthest aisle the order touches and return, so that distance is set by the highest aisle number on the list rather than by how many aisles there are in total.
Those two behave very differently as the list grows. In-aisle travel per line falls as picks concentrate, because the walk to the back gets shared across more picks. Cross-aisle travel rises as the list spreads, because the furthest aisle keeps moving further away. The whole of slotting is an attempt to push work from the second bucket into the first.
Why the same order costs twice as much
Run the defaults and a twelve line order scattered across twelve aisles touches around eight of them, which means eight separate trips down and back with an average of one and a half picks each. In-aisle travel is high per pick because almost every trip is a walk to the back of the aisle to collect one or two items.
Confine the same twelve lines to three aisles and the picker touches essentially all three, with four picks in each. Each trip down and back now serves four picks instead of one and a half, and the cross-aisle distance collapses because the furthest aisle is the third rather than the eleventh. The total is somewhere near half.
Nothing about the order changed. The same twelve items were picked, in the same quantities, by the same person. The difference was decided months earlier, when somebody chose where those items live.
What moves the number, in order of size
| Lever | What it does to travel | What it costs |
|---|---|---|
| Slotting fast movers together | Large — fewer aisles touched, more picks per trip | A move plan and the labor to execute it |
| Batching several orders per trip | Large — one walk serves several orders | Sortation at the end, and system support |
| Shorter aisles, more of them | Moderate — less depth, more crossing | A layout change and lost storage to cross aisles |
| Faster walking | Small and it does not last | Nothing good |
The last row is on the table because it is where attention usually goes first. Travel time is speed divided into distance, and of those two only distance is under anyone's control in a durable way. Pressing on speed produces a short improvement, then fatigue, then errors, then the improvement goes away and the errors do not.
Reading the day figure honestly
The daily total is the one that changes decisions, because per-order numbers are too small to argue with. Six or seven walking hours a day across a picking team, at a fully loaded labor rate, annualises into a figure that comfortably justifies a slotting project — and slotting is one of the few warehouse improvements that costs labor and a plan rather than capital.
Be careful about what the saving actually is, though. Cutting travel by thirty percent does not remove thirty percent of the payroll; it removes thirty percent of the walking, which is a share of the day, and it only becomes money if the freed hours go into more orders or fewer hours scheduled. If neither happens, the improvement is real and invisible. Deciding in advance which of the two it will be is the part that makes the project worth doing. The slotting and ABC placement calculator works out which items to move, and the pallet rack layout calculator covers the aisle geometry those items sit in.
Questions people ask
How much of a picker's day is spent walking?
On the defaults here it is around sixty percent of order time, and figures in that region are common in manual pick operations with modest lines per order. It is not a universal number and it moves a long way with the pick list. A one line order is almost entirely travel, because the walk happens regardless and there is only one pick to spread it over. A forty line order concentrated in a few aisles can be mostly picking. If you want your own figure rather than a plausible one, time a dozen orders with a stopwatch split between moving and picking, and use that.
Does this model batch picking?
No, and batch picking is usually the single largest change available. Picking several orders in one pass means one walk serves many orders, so travel per order falls roughly in proportion to the batch size until congestion and cart capacity stop it. What it adds is a sortation step afterwards, either at a station or into a divided cart, and that step has its own error modes. To approximate a batch here, enter the combined line count of the batch as the lines per order and read the result as the cost of the whole batch rather than of one order.
Should aisles be long or short?
Short aisles cut the walk to the back and cost storage, because every cross aisle is floor that holds nothing. Long aisles store more and make every visit expensive. Which way to lean depends on lines per aisle visit: with one or two picks per aisle, the walk to the back dominates and shorter aisles help; with many picks per aisle, the depth gets amortised and long aisles are fine. Change the aisle length here and watch the in-aisle line rather than the total, since that is the part the geometry controls.
Is the random assumption reasonable?
It is a deliberately unflattering base case rather than a description of your operation. Real demand is concentrated, so a real order touches fewer aisles than the random model predicts, which means the base figure here is closer to a worst case than an average. That is useful for the comparison it is built for: the gap between the random case and the concentrated case brackets what slotting can be worth. If you know your own average aisles per order, enter that as the aisle count and you get something much closer to your actual base.
What walking speed should I use?
Lower than you expect. A comfortable unloaded walk on a clear corridor is faster than a picker pushing a loaded cart, stopping at every location, and starting again, and it is the starting and stopping rather than the top speed that sets the effective figure. The default here is deliberately conservative. If you want to measure it, time a known distance during an actual pick round rather than walking it yourself, because the two are not the same activity.