Spreadsheet Requisitions and Duplicate Suppliers
You are a procurement analyst at a construction firm with 600 employees and 14 sites. Requisitions arrive by email and spreadsheet. There are 4,200 active supplier records, including many duplicates with slightly different names and bank details. About 38% of invoices are matched manually, average approval takes nine days, and maverick spend is estimated at 25%. One site recently bought safety gloves from a non-approved supplier at 30% above the catalogue price because the approved supplier's details could not be found. The CFO wants to buy an e-procurement system but the budget is limited and the last IT project overran. Site managers say the current process is slow and they will keep using email if a new system adds clicks. Finance wants better control; operations want speed. You have been asked to propose a practical improvement plan before any software is selected.
1. Diagnose the root causes of maverick spend and slow approvals, distinguishing data, process and system issues.
Read the worked answer
Diagnosis:
- Data: 4,200 supplier records with duplicates is a master-data problem. It makes it hard to find the approved supplier, creates duplicate payments and weakens spend analysis. Bank-detail duplicates are also a fraud risk.
- Process: email and spreadsheet requisitions have no mandatory fields, no validation and no audit trail. Approval takes nine days because approvers are unclear or absent; users bypass the process to get speed.
- System: there may be no single source of approved suppliers, no catalogue and no automated three-way match. Manual invoice matching at 38% is a symptom of missing PO and receipt data, not just a finance workload.
- Behaviour: site managers are not deliberately maverick; they respond to a process that is slower than email. The safety-glove example shows that poor data drives higher prices.
- Technology will not fix this alone. Buying a system before cleaning data and simplifying approvals is likely to automate the chaos and repeat the last IT overrun.
2. Recommend a prioritised sequence of workflow and system changes for the next 12 months.
Read the worked answer
Prioritised plan:
- Months 1-2: clean the supplier master. Deduplicate by legal name, tax ID and bank details; freeze new supplier creation; require procurement and finance approval. This is low-cost and high-impact.
- Months 2-3: introduce a mandatory purchase requisition form with a valid supplier, budget code, delivery date and approval threshold. Start with a simple email-to-workflow tool if the ERP cannot be configured quickly.
- Months 3-5: build a catalogue for the top 20% of spend categories, starting with safety equipment and consumables. Negotiate catalogue prices and make the approved supplier easy to find.
- Months 5-8: configure approval workflows with delegated limits, automatic reminders and mobile approval. Aim to cut approval time from nine days to three.
- Months 8-12: implement PO-based three-way matching and then select an e-procurement module only if the process is stable. Pilot at two sites, measure adoption and fix pain points before rolling out to all 14.
3. Define measures that would show whether the changes are working.
Read the worked answer
Measures:
- PO compliance: percentage of spend with a valid PO, target above 90% within nine months; maverick spend, target below 10%.
- Master data quality: number of active suppliers; duplicate records; percentage of suppliers with verified bank details.
- Process speed: average requisition-to-approval time; percentage approved within two days; invoice exception rate and manual match rate.
- Commercial outcome: catalogue coverage of spend; price variance against catalogue; savings from rationalised suppliers.
- User adoption: active users per site; percentage of requisitions through the new channel; helpdesk tickets.
- Control: number of retrospective POs and emergency purchases; audit findings. Use a baseline before changes and report monthly to the CFO and site managers. Do not measure only system licence usage; that would miss the process and data problems.