Client
Foresight Metal Engineering
Role
Analysis, design & implementation
Discipline
Warehouse operations

Nothing in the warehouse had an address

What was achieved

Cost savings
95%
Reduction in duplicate orders post inventory tracking
Process improved
63%
Reduction in time to locate an item post addressing
Cost avoided
27%
Reduction in inventory held, after clearing non-moving stock

The warehouse had no addressing system and no stock record. One person knew where things were, and orders were duplicated without him.

What I personally did on this

Owned
Operations analysis · the addressing design · inventory control
Built
The addressing scheme and the tracking that ran on it
Led
Implementation on the floor

A warehouse that could be described

What existed before
  • No SOP and no flowchart for inbound or outbound. Neither had ever been written down
  • No warehouse layout. Racks and pallets sat wherever was convenient
  • No location labelling. Racks and storage unidentified, so locating goods meant asking a person
  • No inventory tracking. A physical check before every delivery order, or memory — which caused ordering of the same item
  • Shipment tracking in AirTable, updated by hand per delivery, with lost documents causing multiple deliveries
What was delivered
  • Inbound and outbound process flowcharts, drafted from observation including the exception paths
  • A warehouse blueprint — 3D concept, measured layout, then final racking plan across two levels
  • A location ID schema applied to every rack position in the building
  • A macro-driven barcode generator with a three-state register that refuses duplicates
  • A new SOP plus training: scan the location on put-away, scan out when preparing a delivery order

The process nobody had ever written down

There was no SOP and no flowchart for either direction. Both of these were drafted from watching the floor, purely so there was a current state to analyse.

Inbound process flowchart with decision branches
Inbound — supplier delivery order checked against physical goods, GRN raised, then the branch that decides warehouse storage or immediate packing
Outbound process flowchart with decision branches
Outbound — picking, the stock-unavailable branch that triggers a purchase order, self-collection, scheduled delivery and proof of delivery
Before — where it broke
Sales prepares the customer delivery order
⚠ Bottleneck Operators check the racks, or remember no location record · no stock record · if wrong, a purchase order is raised for stock already owned
Physical order clipped to a board, AirTable updated by hand
After
Goods received against the supplier order
✓ Addressed Scanned into a location on put-away unit · level and zone · column and row
✓ Tracked Scanned out when the delivery order is prepared movement recorded, not recalled

What shaped the decisions

The finding the whole project turns on

The company reported four problems. They were one problem with four symptoms. A messy floor, miscommunication, duplicate ordering and duplicate deliveries all trace back to the same absence: nothing in the warehouse had a name two people could agree on.

The scope, plainly

Every item on that list is mine, end to end. I mapped the process, ran the data, drew the layout, designed the addressing schema, built the tool, wrote the SOP and delivered the training. There was no team to hand any of it to.

What they needed, and why it mattered

Five requirements. Each one has a business consequence attached, and each consequence shows up in the numbers at the top of this page.

  1. 01
    Write down the process before changing it
    No SOP or flowchart existed for either inbound or outbound. Both were drafted from observation, covering the exception paths as well as the happy path.
    You cannot improve a process nobody has described. Mapping it was the diagnostic, and it is what turned four complaints into one root cause.
    Derived requirement · the blocker to everything else
  2. 02
    Give every position an address that reads like directions
    Warehouse unit, then level and zone, then column and row, at a fixed character count across every location in the building.
    An address a stranger can follow is what removes the dependency on the person who put it there — which is the 63%.
    Derived requirement
  3. 03
    Decide what to keep on data, not on opinion
    Twelve months of movement pulled and pivoted by product code, week, total quantity and delivery-order frequency, then converted to cumulative sales and frequency percentages.
    Disposal is a political conversation in a family business. Handing management a ranked list makes it a decision instead of an argument, and it ran alongside stock take so no extra downtime was needed.
    Client-specified · management made the call
  4. 04
    Deliver it without stopping the warehouse
    A twenty-person company living on daily transactions could not pause to be reorganised. Everything ran in phases around live operations.
    Then Singapore's Circuit Breaker closed the site mid-project and reopened it at reduced headcount. Phasing is the only reason the work survived that rather than being abandoned.
    Constraint
  5. 05
    Make it survive the people who did not want it
    Multiple owners who did not share the same direction, long-tenured staff who saw no need to change, and operators who refused the new SOP outright.
    A process only its author follows is not a process. Repeated meetings and training were the deliverable as much as the schema was.
    Constraint · the real difficulty of the project

Five phases, each one producing what the next one needed

Not a tidy-up. The order could not be shuffled, and Set in Order needed four phases of its own before a single label went on a rack.

01Sort — what should not be in the building at all?
Raw twelve-month movement data extract
Twelve months of movement pulled from the system, product by product, week by week
Pivot of movement data by product code, weeks, sum and frequency
Pivoted to product code against total quantity and delivery-order frequency — volume and how often, separated
Cumulative sales and frequency percentage analysis with slow movers highlighted
Converted to cumulative sales % and frequency %, with the slow movers highlighted for management's disposal decision
What it showed

Sorting a warehouse on opinion produces an argument. Sorting it on twelve months of movement produces a ranked list, and management makes the call rather than defending it.

Phase two ran alongside the stock take, so dead stock was identified and live stock counted and located in a single pass. No extra downtime.

This is the analysis the 27% comes from. The warehouse was full, and the working assumption was that the business had outgrown it and needed a second unit. The ranked list said otherwise: a substantial share of the floor was held by items with almost no movement across the whole twelve months. Clearing them reduced inventory held by 27% and released the space that had supposedly run out. The second unit was never rented.

Worth being precise about what that means commercially. Twenty-seven percent less stock is cash the business is no longer holding on a shelf, and it recurs. Avoiding the second unit is rent, deposit, fit-out and the staff time to run two sites, and that avoidance came from a spreadsheet rather than a capital decision.

02Set in Order — where should everything go?
3D concept drawing of the proposed warehouse layout
Phase 1 — a 3D concept, deliberately not to scale. Its job was to get agreement, and it had to allow for a planned mezzanine
Measured warehouse layout plan
Phase 2 — measured. Racking placed against product weight, the company's safety standard and ease of manoeuvring
Final layout pairing the 3D isometric with the 2D plan across two levels
The signed-off version — 3D isometric tied back to the plan view with callouts, across both levels including the mezzanine
Why it took three drawings

The concept exists to get a decision out of several owners who did not share a direction. The measured version exists to be built from. Skipping the first means arguing about millimetres before anyone has agreed on the shape.

03The addressing plan — every position in the building
Warehouse location plan with every rack range assigned an ID across two levels
Not a schema on paper — the applied plan. Every rack range assigned across L1 and L2, with stairs, toilet and entrance mapped
The design decision inside the ID

L427 – 1A – 0101 is warehouse unit, then level and zone, then column and row — the order a person physically walks it. The code tells you which building, which floor, which aisle, which shelf, in that sequence.

The character count is standardised across every location, which is what makes it sort, scan and print predictably at volume.

04The tool that applied it
Excel barcode generator showing three register states and a colour legend
Built in Excel. Three states — Please Register, Registered Successfully, Barcode Exists, Do Not Register — with a colour legend for user input, information, success, alert and error
Generated location barcode applied to a warehouse rack
Generated, registered, printed and applied to the rack
Why it needed states at all

Mass-producing IDs by hand loses track of what has been issued, and a duplicate address reintroduces the exact ambiguity the whole project existed to remove. The register refuses a code that already exists instead of silently overwriting it.

The colour legend is there because the people using it did not want the change. A tool that tells you plainly what state you are in is a tool that gets used.

05Standardise and sustain — making it stick
The new standard operating procedure workflow
The new SOP — scan the location on put-away, scan out of it when preparing a delivery order
5S ownership board with personnel, zone and abnormality feedback
The 5S ownership board: who is responsible for which zone, with an abnormality and feedback sheet attached. Proposed and still under review at handover
What was still open

The first cycle count found discrepancies and they were unresolved when the engagement ended. The ownership board was accepted in principle and never finished. Both are here because leaving them out would make the rest less believable.

What it actually looked like

Warehouse floor before the project, goods stored without structure
Before — items of any type placed anywhere, mixed, with no accountability
Warehouse floor before the project, racking without orientation
Before — no racking orientation and no structure
Warehouse before and after the implementation
After — materials well kept, positions addressed, movement tracked
The three things that nearly stopped it

Multiple owners not aligned on direction, with the strongest resistance from the longest-serving staff. No downtime available in a twenty-person company living on daily transactions. And Circuit Breaker, which closed the site mid-project, then reopened it at limited headcount and did not give the timeline back.

What produced each result

Three deliverables, and each one maps to a number at the top of this page.

Produced the 95%

Scanned inventory tracking, in and out

Scan the location on put-away, scan out when preparing a delivery order. Stock stopped being a memory, which is what stops the company buying what it already owns.

Produced the 63%

A measured layout and an address that reads like directions

Racking placed against weight, safety and manoeuvring, then every position given an ID that walks you to it: unit, level and zone, column and row.

Produced the 27%

The stock classification, and the disposal it justified

Ranking every item by movement separated what the warehouse was actively using from what it was only storing. That distinction is what turned “we need more space” into a disposal decision management could sign, and it ran in the same pass as the count so it cost no extra downtime. The barcode generator and the addressing register, which made every position identifiable from zero, are what allowed the count to be done at that granularity in the first place.

And the one nobody asked for

The process flowcharts themselves. Inbound and outbound had never been documented. They were drawn purely to create a current state that could be analysed, and became the company's first written description of how it actually operates.

Part two of two

How it was actually done

Everything above is the business read. Everything below is the work itself — the data, the drawings, the addressing plan, the tool and the floor.

!Before you open this

Everything above is the bottom line. If that is what you came for, you already have it.

What follows is the full working: 16 diagrams and 6 sections of working, the analysis behind each decision, and why it went that way instead of the obvious way. It is long on purpose. It is written to be checked, not skimmed.

Only wanted the overview? Stop here. You will not miss a single result — every number is already above this line.