Procurement Anomalies
Split-Order Clustering
Detects orders split to stay under approval or competitive-bidding thresholds.
Why Split Orders Are Irregular
Splitting a requirement into smaller orders to stay below a quotation or competitive-bidding threshold deliberately evades SCM prescripts and is treated as irregular. It defeats the fairness and competitiveness that section 217 of the Constitution demands. AuditPro Core clusters related orders by supplier, timing and description so deliberate splitting is exposed even when each individual order looks compliant.
The Numbers
AuditPro Core renders this view from your tenant's live, tamper-evident records. The figures below are illustrative sample data.
Split Clusters
143
▲ 18
Value in Clusters
R 54.2m
near thresholds
Just Below R500k
61
clusters
Vendors Involved
88
distinct
Split Clusters by Threshold Band
Top Split Clusters
| Vendor | Orders | Window (days) | Total (R'000) |
|---|---|---|---|
| Vendor 0471 | 5 | 3 | 2440 |
| Vendor 0312 | 4 | 2 | 1960 |
| Vendor 0588 | 6 | 5 | 2870 |
| Vendor 0149 | 3 | 1 | 1480 |
| Vendor 0623 | 4 | 4 | 1990 |
Figures shown are illustrative sample data for demonstration. AuditPro Core renders these views from your own tenant's live, tamper-evident records.
How Splitting Works
A purchase that should trigger three quotes or a tender is broken into pieces that each fall under the threshold. Each piece appears compliant in isolation, which is precisely why split detection requires looking across transactions.
Clustering Signals
Multiple orders to the same supplier within a short window, for similar goods, summing to just above a threshold, are the classic fingerprint. The closer the cluster sits to the threshold boundary, the stronger the suspicion.
Intent and Defensibility
Not every cluster is deliberate; genuine operational batching can resemble splitting. The control surfaces candidates for explanation rather than presuming guilt.
How AuditPro Core Bridges the Gap
- Order clustering: related orders are grouped by supplier, timing and description to reveal aggregate values that breach thresholds.
- Threshold mapping: clusters are tested against the relevant quotation and bidding limits to highlight near-boundary patterns.
- Exception workflow: suspected splits route to SCM for justification or escalation to the irregular register.
- Traceability to source: each cluster links to its component orders for evidence-backed follow-up.
Key Takeaways
- Splitting to dodge a threshold is irregular even when each order looks clean.
- The fingerprint is same supplier, short window, similar goods, sum above threshold.
- Clusters just over a threshold deserve the closest scrutiny.
- Treat detections as questions to answer, not automatic findings.
See This on Your Own Data
AuditPro Core renders this dashboard from your tenant's live, tamper-evident records — every figure traceable to source.