Gold Trading Operational Transformation
- Industry
- Commodity trading (gold)
- Client
- A Lagos-based gold trading business
- Engagement
- Independent consulting: analysis, design, implementation
- Duration
- 3 weeks
- Primary outcome
- Daily reconciliation reduced to under 30 minutes
- Status
- Live. In daily use since handover.
Business context
The business had grown past the processes that originally supported it. Orders, fulfilment, sales, receivables, and payables were each tracked in separate physical books, and the records only became trustworthy after a long manual reconciliation at the end of each day. A deeper monthly reconciliation could take almost a full working day. The process worked because the owner personally remembered how everything connected. Nothing was documented, and every increase in transaction volume made the daily coordination cost heavier.
The objective was not to automate the existing work. It was to redesign how work moved through the business, and only then decide what deserved automation.
Discovery
The engagement began by documenting how work actually happened, not how anyone assumed it happened. Each operational activity was mapped from the moment an order arrived through to end-of-day reconciliation. For every step, the following were recorded: inputs, outputs, decision points, manual interventions, dependencies, failure points, and the workarounds the owner had built over time.
Only after the current operating model was fully understood were any changes proposed.
Current workflow
Before
- Order arrives by WhatsApp, Telegram, Instagram or TikTok DM, or a referral walk-in
- Employee records the order in a physical book
- Employee confirms the sale and payment details to the client in the DM thread
- Client pays by bank transfer
- Owner confirms the transfer against the bank account
- Order processed and fulfilled
- Fulfilment logged in the books
- End of day: owner reconciles bank statements against the order and log books
Multiple manual handoffs meant the same information was entered more than once, and the numbers could not be trusted until the owner had personally reconciled every record.
What discovery found
- Multiple sources of truth. Operational information lived in several disconnected physical books.
- Duplicate entry. The same order and payment details were entered into more than one record.
- Manual reconciliation. Records required extensive cross-checking before daily numbers could be trusted.
- Owner dependency. The process existed in the owner's memory, not in documentation.
- No validation. Errors were discovered during reconciliation, after they had already spread across records.
- Poor scaling. Every additional transaction increased manual coordination time.
Design decisions
Decision
One operational record.
Why
Information should exist once. Duplicate entry was the root cause of most reconciliation work.
Decision
Validation before reporting.
Why
Errors should be caught before numbers are trusted, not discovered afterwards during reconciliation.
Decision
Automate repetition, keep judgment.
Why
Calculations and cross-checks are deterministic and were automated. Approval of critical financial actions stayed with the owner.
Decision
Keep order intake on the channels the business already used.
Why
Adoption mattered more than sophistication. Orders kept arriving through the same WhatsApp, Telegram, and social DMs; the system changed what happened after they arrived.
Redesigned workflow
After
- Customer order received
- Recorded once in the operational system
- Validation rules applied
- Invoice and cash tracking generated from the same record
- Reconciliation view
- Daily position
Implementation
The redesigned workflow was introduced in phases rather than as an overnight replacement, so the business never stopped operating during the change. Documentation was written alongside implementation, not after it. The owner was trained on the new workflow before handover, and the engagement closed only once the system was being operated independently.
Results
What this engagement demonstrates
Process discovery. Workflow mapping. Requirements gathering. Current-state and future-state design. Decision logic. Stakeholder training and handover. Impact measurement against a baseline.
Reflection
Redesigning the workflow produced more value than automating the existing process would have. Once operational responsibilities, decision points, and information flow were clear, the technology choices became straightforward, and the appropriate level of automation became obvious. The system has run independently since handover because the process was designed to outlive the engagement.