Cluster picking in D365 WMS: one walk, many orders
How cluster picking in Dynamics 365 Warehouse Management lets one picker fill several orders in a single trip, when it pays off, and the setup decisions that make or break it.
Sample content. This is a placeholder post to preview the layout. It is not indexed by search engines.
Picture a warehouse that ships a few hundred small orders a day. Picking them one at a time means the same aisles get walked over and over, often for the same fast-moving items. Cluster picking fixes that: one worker takes a cart with several totes, picks for several orders in one route, and drops each item into the right tote as they go.
In Dynamics 365 Warehouse Management, cluster picking is standard functionality. You don't need custom code, but you do need to make a few setup decisions deliberately.
See it run
The animation below runs one wave from dock to dock. Watch the status card at the Wave step: three sales orders are grouped into one wave and released as a single cluster-pick job, and the cart then picks all of them on one route. Switch between the two layouts to see how a forward pick zone with replenishment shortens that route further.
When cluster picking pays off
Cluster picking earns its keep when travel is the biggest part of picking time:
- Many small orders. E-commerce, spare parts and B2B replenishment orders with a handful of lines each.
- Overlapping items. If several orders on the cart need the same item, the worker picks it once per location and sorts it into positions.
- Each and case picking from shelving or a forward pick zone. These are the places a cart can actually reach.
It helps much less when orders are big enough to fill a cart on their own, or when you pick full pallets. Those flows are better served by order picking or by a different work template.
What D365 needs
Microsoft's setup has five parts, and each one is a decision rather than a checkbox:
| Component | What it decides |
|---|---|
| Cluster profile | How many positions a cluster has, how they're named, when the cluster breaks, and how picks are sorted and verified |
| Work templates | How the picking work is created, typically one work ID per order so each one gets its own position |
| Location directives | Where to pick from and where the picked goods go |
| Mobile device menu item | A menu item directed by cluster picking and linked to the cluster profile |
| Warehouse management parameters | The number sequence for cluster IDs, if the system generates them |
The cluster profile decisions that matter
Cluster profiles live under Warehouse management > Setup > Mobile device > Cluster profiles. These are the settings we see cause most of the trouble in testing:
- Number of positions. Match the physical cart. If the cart holds six totes, a profile with eight positions creates work the floor can't carry.
- Position names. Numeric or alphabetical names that match the labels on the totes. Workers will scan or read them hundreds of times a day.
- Break cluster at. Decides when the cluster is finished. Breaking at the put step keeps the cluster together until goods reach staging or pack.
- Sort verification type. A position scan when sorting into a tote catches the classic error of the right item going into the wrong order's tote.
- Cluster sorting. Sort by location so the route follows the aisle layout instead of zig-zagging.
In version 10.0.48 and later, the profile also has a cluster picking strategy. Process by location picks the total for all positions at a location in one scan and then asks the worker to sort it. Process by position handles each position separately, for all items or only batch- and serial-tracked ones. If you track batches or serials below location level, test this setting early, because it changes the mobile flow your trainers need to teach.
On the floor
With manual clusters, the worker starts on the mobile device by assigning work IDs to positions, one order per tote. The device then walks them location by location: pick the quantity shown, then sort it into each position and confirm with a scan.
Two behaviours are worth briefing your supervisors on before go-live:
- Once work is in a cluster, only cluster picking can process it. If an order lands in the wrong cluster, the cluster has to be broken and rebuilt.
- Clusters can be passed to another worker, which helps at shift change or when someone is pulled off the floor.
If you don't want workers building clusters by hand, system-directed cluster picking lets D365 build the clusters itself from released work.
Where to start
Begin with one cart type, one cluster profile and one work template for your highest-volume small-order flow. Get that stable in testing with real tote labels and real scanners, then add profiles for other order types.
Further reading on Microsoft Learn: Set up cluster picking and System-directed cluster picking.
