Item level picking and PTL Segregation
INCREASING ITEMS PICKED PER HOUR BY 17.2%
OVERVIEW
Quick commerce warehouses store inventory in bulk and serve as fulfillment centers for dark stores.
To optimize the dispatch process from the warehouse to dark stores, this project focused on improving the item picking and sorting process. We aimed to introduce Item-Level Picking to optimize pickers' pick paths and implemented a Put-to-Light (PTL) system for faster and more accurate store-wise item sorting.
ROLE
Product Designer
Led the design process by understanding user needs, mapping workflows, crafting interface, and collaborating with PMs, engineers, and operations team to optimize warehouse processes.
USERS
Picker, Transporter, Segregator
IMAPCT
17.2%
Increase in items picked per hour per picker, improving overall picking efficiency and optimizing pick path
25%
Reduction in segregation errors eliminating the need for additional QC step due to less picking errors
60%
Increase in overall productivity, driven by combined effect of Item level picking and PTL based segregation
Warehouse (FC) process overview
Problem discovery and understanding current workflow
As part of the discovery process, I regularly visit the warehouse to understand the users, their behavior, and the warehouse processes.
This problem area was discovered while observing a few pickers, and when I tried picking items using the users' handheld devices to better understand their workflow. Through this hands-on experience, I identified several areas where the current picking process be improved.
Process overview: Store-level picking
Previously, store level pick lists were created, a picker at once would pick items for a single store as a part of one pick list which resulted in picking multiple SKUs (items) in smaller quantities from multiple locations.
These pick lists are created on store + dispatch time level and each pick list can only have items kept on the same floor.
Picking items from multiple locations
Dropping picked items in crates
Dropping all crates through spiral chute
A transporter moving crates from spiral chute to QC station, where all quantities are checked against a pick list.
Sorter sorting all QC crates of a store on a pallet and placing the pallet on the dispatch lane (All crates in the image are sorted)
STORE LEVEL PICK LIST CHALLENGES
Pick Path
Lack of pick path optimization leads to increased physical effort as pickers repeatedly move between locations.
Picking Time
Due to pick path not being optimized, picking time also increased to complete a all pick lists for a store
Picking Errors
Increase in picking errors due to picking multiple SKUs from multiple locations at a single time
QC Dependency
High dependency on QC to check any errors in picked quantity
The above image is of a store level pick list task. The user is expected to pick 38 SKUs and 44 Units in a single pick list. This would most likely require the picker to go to at least 25 - 35 different locations to pick 44 units.
PROBLEM STATEMENT
The current store-level picking process in warehouse increases the physical effort of pickers resulting in slower and inaccurate picking of items.
This fragmented workflow negatively impacts overall fulfillment speed, accuracy, and downstream QC efforts, affecting the timely fulfillment of orders.
PROPOSED SOLUTION
Enable item-level picking to collect items for multiple stores in one go, followed by store-wise segregation and quantity checks using PTL (put to light) or pigeonhole racks.
This aims to improve overall efficiency while eliminating the need for traditional QC processes.
PTL AND PIGEONHOLE SETUP
Pigeonhole rack system created at Bangalore warehouse to test the system before getting PTL systems
PTL (Put to Light) uses light displays, indicating the specific locations where items need to be placed or "put" after being picked. (The above image is only for reference)
USERS AND THEIR WORKFLOW
Picker
The picking process on the HHD device remains unchanged, but the picker’s workflow is simplified as they now need to pick SKUs in bulk from only 1 to 3 locations in a single pick list.
Transporter
A transporter identifies crate types at the spiral chute exit and moves them to the correct station. Earlier, all crates went for QC, but with item-level pick lists, item-level crates will now go to segregation station (PTL) instead.
Segregator
With store level pick lists, there was no need for a segregator, instead after picking, crates were sent for QC. Now the segregator needs to segregate items for different stores eliminating the need for QC process.
Starting with the designs and ensuring the app closely mimics actual user behavior on the ground, taking into account operational constraints in the warehouse.
WORKFLOW FOR THE TRANSPORTER
All PTL flow entry points
Scan IPST zone
Scan all crates
Assigning a IPST Zone
While entering this flow, the transporter is at the exit of the spiral chute.
The transporter has to assign a IPST zone (spiral chute exit) to themselves and scan all the crates that the picker has dropped through the spiral chute.
User scans Store Level Crate
User scans Item Level Crate
Scanning Crates
As the user scans crates, they'll see instruction to move the crate.
Store Level Crate: Move to QC Station
Item Level Crate: Move to PTL Station
Process on ground?
On the ground, transporters use two pallets: one for store-level crates and one for item-level. After scanning a crate, they place it on the assigned pallet and move it to its station.
All scanned crates
Why have an interface for the repetitive task of scanning and palletizing crates, especially when a screen adds friction to the flow?
The initial idea was to use color-coded crates (yellow for item-level, blue for store-level) instead of a screen to simplify the task of transporter. However, this was not feasible due to the cost implications and the high volume of crates needed daily, which would have blocked picking if procurement fell short due to continuous movement of crate in and out of warehouse.
UI EXPLORATION
Explored a few options to highlight crate type and information.
Explored a few options considering some design goals:
-
The interface shouldn't block the user
-
Instruction to move crate and "where to move crate" should be upfront and visible
-
Information has to be color coded considering it will help users recognize crate type without reading the information on screen.
WORKFLOW FOR THE SEGREGATOR - PTL INTEGRATION
Once the crates are transported to the segregation (PTL) area, the segregator has to pick any crate and start segregating items based on store. User can start with any item they see first in the crate.
User scans any PTL station
Scans any crate kept in the PTL area
Scans any item (SKU) from the scanned crate
User selects item variant
Once the user selects the item variant, they need to drop the items into the respective store crates placed in bins on the rack. A segregator maps these crates to bins, and based on the scanned item, the system prompts the user to drop the required quantity.
The system already knows the quantity needed for each store, so when an item is scanned, it displays instructions to drop the correct quantity into the crate mapped to that store in the backend.
UI APPROACH 1
UI APPROACH 2
FLOW APPROACH 1
Segregator will scan and drop one unit of an item at a time.
This might eliminate the need for QC process as the segregator will be dropping one unit at a time resulting in accurate dropping of item
FLOW APPROACH 2
Segregator will scan one SKU and drop all units of that item in relevant drop crates.
In this approach productivity might be higher since multiple quantity will be dropped at once but there are high chances of errors
Scan crate mapped to the bin
If the drop crate is full, user needs to unmap and close the crate
EDGE CASES
If PTL station is already assigned to someone else
If crates mapped to all bins are full and no location is suggested
Any wrong crate is scanned which is not mapped to the bin
If the user tries to close ILP crate without segregating all items
Excess Item
User scans any excess item or adds excess quantity which was not in the picklist but might be present in the crate due to picking error. In this case the user will be asked to drop the excess item in an excess crate.
SOME MORE EXPLORATIONS
I also explored installing TV panels with each PTL setup to display bin IDs on-screen instead of on handheld devices, with scanners mounted alongside the PTL, removing the dependency on HHDs
But again this was not feasible due to the cost and we wanted to build conviction on item level picking and PTL setup first before investing a lot on new infrastructure.
TESTING AND IMPACT
We successfully launched Item level picking for some specific items and pigeonhole segregation setup for 2 warehouses and observed significant improvement in picking speed and accuracy.
Testing of this setup was done for only one approach where the user is asked to input quantity, going forward we are planning to test other approach too without the dependency on QC.
Going forward we are planning to invest in advance PTL setup which will minimize the dependency on the hand held device
Some metrics that proved the effectiveness of introducing Item level picking and PTL segregation
17.2%
Increase in items picked per hour per picker, improving overall picking efficiency and optimizing pick path
25%
Reduction in segregation errors eliminating the need for additional QC step due to less picking errors
60%
Increase in overall productivity, driven by combined effect of Item level picking and PTL based segregation
After the success of the pigeonhole setup, we are now testing PTL with a TV display to eliminate dependency on handheld device and ease item handling.
