Walking and driving between locations is work that produces nothing, and in a picking operation it typically dominates task time.
Estimating it without new hardware
Task sequences give location pairs. Each consecutive pick implies a movement between two known locations.
A distance matrix for the facility converts pairs into distance. Building one is a day's work with a floor plan.
Sum by task, by shift, by product, which gives a travel profile without any positioning system.
Accurate enough for decisions, and it costs nothing beyond the join.
Positioning systems refine it and are rarely necessary to identify the big losses.
What the profile reveals
Fast-moving items in distant locations, which is the classic slotting failure and the easiest win.
Product families split across the facility, so a typical order requires crossing it.
Congestion points where aisles are shared.
Return trips to a start point that could be eliminated.
Search time, appearing as unexplained gaps between scans in a known location.
Slotting
The main lever, and it is an ongoing activity rather than a project.
Rank items by pick frequency, not by sales value, which is a different ranking.
Place high-frequency items closest to the despatch or consolidation point.
Group items that appear together in orders, which reduces travel more than frequency ranking alone.
Account for size and weight, since heavy items belong at accessible heights regardless of frequency.
Re-slot on a cycle, because demand changes and a slotting done once decays.
Measure the travel before and after, which is what makes the case for doing it again.
Batching and routing
Batch picking — collecting several orders in one trip — reduces travel substantially and increases sortation work.
Zone picking removes long trips and adds consolidation.
Route optimisation within a pick list, which is a solved problem and frequently not implemented.
Each trades one cost for another, and the data tells you which trade is favourable in your facility. Copying another operation's choice without the analysis is how facilities end up with a method that suits a different mix.
Congestion
Distinct from distance and frequently more costly.
Aisles used by several people simultaneously produce waiting that looks like slow picking.
Equipment sharing in narrow aisles.
Bidirectional traffic in aisles designed for one direction.
Detect it by looking for locations where task times are consistently longer than distance predicts, at particular hours.
Fix it with routing rules, one-way flows, or zone assignment before considering physical changes.
What to report
Travel as a proportion of task time.
Distance per order line, trended.
The ten locations contributing most travel, which is the slotting work queue.
Congestion hotspots by location and hour.
All computable from data most facilities already hold, which makes this the highest-return first analysis in a picking operation.
The re-slotting cycle
Slotting decays because demand moves, and a one-off exercise loses its value within a year.
Re-rank pick frequency quarterly, which is a query.
Identify the items whose rank has moved most.
Move the top movers only, which is a manageable amount of physical work.
Do it in the trough, not at peak.
Measure travel before and after, which is what justifies doing it again.
Automate the ranking so the exercise is a decision rather than a project each time.
Affinity, not just frequency
Frequency ranking alone misses the larger saving.
Items that appear together in orders should be stored together, even where individual frequency is moderate.
Compute affinity from order line co-occurrence, which is a straightforward query.
Place high-affinity groups in one zone, which shortens the whole pick path rather than one leg of it.
Check against physical constraints: weight, size, temperature, segregation rules.
Affinity slotting frequently beats frequency slotting in operations with multi-line orders, and almost nobody computes it.