Warehouse slotting and ABC/XYZ analysis - how goods placement speeds up picking

Q: How often does slotting need to be repeated?

Review class A every quarter, do a full re-slotting once a year, plus seasonal rotations if the assortment calls for them. The range turns over at 20-40% a year, so slotting done once and left to itself stops matching reality after two years. The model that works best is one where small corrections happen continuously (the system suggests better slots at replenishment) while the big reviews have a fixed date on the calendar.

Q: Do you need a dedicated WMS module for slotting, or is Excel enough?

To start, an export of the pick history from the WMS and a spreadsheet are enough - an ABC analysis by pick lines is a single pivot table. Excel stops being enough with dynamic slotting (the system has to suggest slots itself at receipt and replenishment) and with market-basket analysis over a large number of orders. More important than the tool is the data: without a reliable history of scans per location, no module can compute anything.

Q: Where do you start slotting with several thousand SKUs?

With Pareto: count the pick lines from the last 3-6 months, sort descending and take the top 200-300 SKUs - they usually generate 60-70% of all the movement. Check where they sit and relocate the ones that are furthest from the packing area or outside the golden zone. The first 100 moves typically deliver half of the whole possible effect. The rest of the range is tidied up in batches over the following weeks, without stopping operations.

Q: Does slotting make sense in a small warehouse, e.g. 500 m2?

Yes, though the absolute effect is smaller - with short travel paths the saving on metres is more modest. But in small warehouses slotting makes the difference elsewhere: the golden zone (less bending means faster picks and fewer injuries), product families (things bought together sit next to each other) and pick-slot capacity (fewer replenishments). The effort at 500 m2 is two or three days of analysis and relocations - it pays back in the first month.

Andrzej Lenkowski, 26 May 2026

Slotting is the decision about where every SKU physically sits in the warehouse. It sounds trivial - the goods have to go somewhere, after all. The catch is that in a typical warehouse that decision gets made once, at the first receipt, on the "there was a free spot" principle. Two years later the item ordered twenty times a day is sitting at the far end of the hall on the third level, while a pallet nobody has touched since March blocks the best location next to the packing area. Operators walk, and walk, and walk - and every metre covered is time you can measure and nobody gets back.

Why travel eats half the picking time

In a classic person-to-goods warehouse the operator spends 50-60% of the shift moving between locations. Not scanning, not packing, not counting - walking. Measurements from real deployments show 8-12 kilometres a day per operator in a mid-sized warehouse, and in large halls with long aisles it can hit 15. It is the single biggest cost item in picking and, at the same time, the easiest one to cut - because it takes no equipment and no headcount, just moving the stock.

The maths is simple. If an operator does 100 lines a day and walks an average of 40 metres per line, that is 4 kilometres. Trimming the average travel by 30% - a typical result of the first proper slotting exercise - means over a kilometre less walking per person per day. With a team of 10 operators and 250 working days that is more than 3,000 labour hours a year recovered without a single złoty spent on equipment.

ABC analysis - but by pick lines, not by revenue

The classic ABC analysis splits the assortment along the Pareto principle: class A is the ~20% of SKUs that generate ~80% of the movement, B is the next 30% of SKUs with ~15% of the movement, and C is the remaining half of the range that barely moves. So far, so agreed. The devil is in the question of what counts as "movement".

Accounting and sales naturally count revenue. For slotting, that is a trap. A television at 8,000 złoty sold once a week generates more revenue than packing tape sold 60 times a day - but it is the tape the operator walks to 60 times. From a travel standpoint what matters is the number of pick lines (how many times someone physically walked up to a location), not value and not even unit quantity. Slotting done by revenue puts expensive goods next to the packing area instead of frequently picked ones - and the warehouse still walks kilometres, it just looks prettier in the spreadsheet.

The second trap is the data window. An analysis of the last month will catch a seasonal peak or a trough - and either one skews the result. A sensible minimum is 3 months of pick history, with the optimum being 6-12 broken down by season if the assortment is seasonal.

XYZ - the second axis everyone forgets

ABC tells you how often an item is picked. It says nothing about how predictable that movement is. That is what XYZ analysis is for, classifying SKUs by demand variability (the coefficient of variation, i.e. standard deviation divided by the mean):

X - stable demand, variability up to ~20%. Tape, cartons, void fill, catalogue bestsellers. Can be planned well ahead.
Y - variable demand, but with a graspable pattern (seasonality, promotions), variability 20-50%.
Z - sporadic, unpredictable demand, variability above 50%. End-of-line stock, spare parts, made-to-order products.

Crossing the two axes gives a nine-class matrix, and the practical weight sits in the extremes. AX is the iron core of the warehouse - high rotation, stable demand; these SKUs get the best locations permanently. AZ is the hardest - the goods move a lot but irregularly, and here a flexible zone assignment works better than a fixed slot. The CZs can safely go up onto the mezzanine or the top levels and be forgotten. Without the XYZ axis every "A" gets treated the same - and the operational difference between AX and AZ is enormous.

Golden zone - physiology, not folklore

The second dimension of slotting is height. The band between 60 and 160 cm off the floor - roughly from the knees to eye level - is the so-called golden zone: the operator picks without bending down and without reaching up. A pick from that band typically takes 3-5 seconds. The same pick from floor level or from 180+ cm takes 8-15 seconds, because now there is a squat, a reach, sometimes a step ladder. Across a few hundred picks a day the difference runs into double-digit minutes, on top of fatigue and injury risk - back strain is the most common category of sick leave among warehouse staff.

Practical rules: class A goods in the golden zone, heavy items (above ~15 kg) between 40 and 80 cm - so they are lifted from hip height, not off the floor and not overhead. Light, slow-moving stock goes to the top of the rack. Fast-moving small items go in bins on flow racks where you have them. And one thing people forget: the golden zone is measured from whatever level the operator is standing on, so it applies on the mezzanine just the same.

More than rotation - what else decides the slot

If pick frequency were the only thing that mattered, slotting would be simple sorting. In practice a handful of extra criteria come in, and they can reshuffle the result:

Product families (affinity). If in 70% of orders customers buy product A together with product B, those two SKUs should sit close to each other - otherwise every such order triggers two separate trips. A market-basket analysis over the order history in the WMS picks up those pairs in an hour.
Cube-per-order index. The ratio of the volume a slot occupies to the number of picks. A bulky item with average rotation should not squat on the best zone, because it "costs" more of it than three fast-moving small items that would fit in the same space.
Weight and pick sequence. Heavy items at the start of the pick path - so they land on the pallet first, on the bottom. Putting the mineral water at the end of the route guarantees crushed crisp boxes on every mixed pallet.
Storage requirements. Chemicals away from food, aerosols in the ADR zone, temperature control, customs zones - these are hard constraints that come before any optimisation.
Slot capacity vs replenishment size. A pick location should hold a multiple of the case pack - ideally 1-2 days of sales. Too small means replenishing three times a day, too large wastes the golden zone.

Static, dynamic and seasonal slotting

Static slotting is a fixed assignment of a SKU to a pick location. Easy to maintain, predictable for people (after a week the operator knows the slots by heart), but rigid - a freshly launched bestseller will sit for months wherever it was first put. It works with a stable assortment of up to a few thousand SKUs.

Dynamic slotting hands the decision to the system: the WMS analyses rotation on the fly and, at replenishment or receipt, suggests ever better slots. The assortment "drifts" toward the fast zones as its rotation grows and drifts away as rotation fades. It demands trust in the system and scanning discipline, because only the WMS knows where the goods are - but with a churning assortment (e-commerce, fashion, electronics) the gap over static is impossible to close.

The middle option, in practice the most common: fixed zones, floating slots. Class A has a reserved area near packing, but the specific slot within the zone is assigned by the system. Add seasonal re-slotting done 2-4 times a year: ahead of barbecue season charcoal and firelighters roll into zone A, in October they go back to the far racks and their place is taken by grave candles. The calendar of these rotations in Polish retail is surprisingly repeatable year to year - you only have to write it down once.

What it actually delivers

The numbers from deployments where slotting was done on WMS data (not by gut feel) line up consistently:

Average pick path shortened by 20-35% - which translates directly into throughput measured in lines per labour hour.
Picking productivity up by 15-25% with no change in headcount or equipment. The upper end applies to warehouses that had never done any slotting before.
Pick errors down by 10-20% - a less tired operator makes fewer mistakes, and a logical layout (similar SKUs not sitting side by side) eliminates the classic "grabbed it off the next shelf" mix-ups.
Shorter queues in the aisles. Spreading class A across several aisles instead of one "golden" one clears the truck jams at peak - a problem that shows up in no report yet can eat 10% of a shift's output.

The cost? With an assortment of up to 5 thousand SKUs the first full slotting is typically 2-4 weeks of analysis and relocations done on the second shift or over weekends. Relocating does not stop the warehouse - you do it zone by zone, a few hundred locations at a time.

How to work it out on the data you already have

Sensible slotting does not need a consultant with an AI licence. It needs the operational history that any working WMS already collects. The minimum data set:

pick-line history per SKU from 3-6 months (date, SKU, location, quantity),
dimensions and weights of the unit and case packs,
a location map with dimensions and levels (the warehouse topology in the WMS),
order history for the market-basket analysis (what ships together).

The procedure fits in four steps: count pick lines per SKU and sort descending; overlay ABC classes by lines and XYZ by variability; map the classes onto warehouse zones (A - closest to packing and in the golden zone, C - the periphery); generate a move list and work through it in batches, starting with the SKUs where the gap between the current slot and the target is largest. The first 100 moves usually deliver half of the whole effect - Pareto works here too.

Common mistakes

Slotting by the manager's intuition. The shift manager "knows" what rotates - and is right in 60-70% of cases. The trouble is the other 30%: SKUs that used to rotate, memories from a season two years ago, personal habits. System data has no sentiment.
A one-off exercise instead of a process. Slotting done in 2023 and never updated is chaos again by 2026 - the assortment turns over at 20-40% a year. The minimum is a quarterly review of class A and a full re-slotting once a year.
Moving everything at once. A grand revolution over a single weekend ends in a week of chaos, because operators hunt for goods in the old slots and replenishments head for locations that no longer exist. Zone by zone, in batches, with the system updated before the physical move - never the other way around.
Ignoring replenishment. Moving a SKU to a smaller pick location without recalculating replenishment frequency can eat the whole gain - what the picker saves on travel, the replenisher gives back in runs from the buffer.
No slack in zone A. A fast zone filled to 100% has no room for anything new. A healthy buffer is 10-15% of slots left empty - room for tomorrow's bestsellers.

Summary

Slotting is the cheapest warehouse optimisation there is: it takes no equipment, no headcount, just data and consistency. The order is always the same - first the ABC analysis by pick lines, then the XYZ axis, then the golden zone and the extra criteria, and finally a re-slotting process written into the calendar, not into good intentions.

In Weaver WMS the rotation analysis is built into the system's reports - ABC classes computed by pick lines over any period, rotation cross-referenced with the current layout, and location suggestions at receipt and replenishment. From the experience of several hundred deployments: warehouses that slot regularly end up, after two years, with picking productivity a quarter higher than those that did it once "for the opening" - on identical equipment and headcount.