In most cases, procurement departments have been tracking suppliers to some degree. The group has a supplier scorecard. They have a quarterly business review. Someone tracks the percentage of parts delivered on-time. But there’s always a lag. The supplier may have had a string of late shipments months earlier. Those late parts may have caused a line to shut down or delayed a shipment to customers.
Continuous monitoring of supplier performance through advanced analytics creates different timing for problems than traditional scorecards.
Instead of looking backward at the end of every quarter, the procurement department now receives continuous monitoring of delivery times, product quality, cost and potential risks across the supply base. And now, the procurement department receives advance notice if a supplier begins to perform poorly.
This article describes examples of how to implement continuous monitoring of supplier performance using advanced analytics. The description will cover the types of data needed for implementation; the metrics that provide insight into how well suppliers are performing; the levels of analytical capabilities available for monitoring supplier performance; and an outline for implementing advanced analytics in a procurement environment.
Limitations of traditional supplier scorecards
Traditional scorecards are a good idea. However, traditional scorecards are not very effective due to how they are developed and used.
- They are too slow. Most data is manually entered from various systems. Therefore, the information displayed on the scorecard represents performance during the previous quarter. Not last week.
- They obscure important trends. A 92% on-time delivery rate may conceal the fact that one particular item is consistently delivered late.
- They are isolated. Information regarding delivery performance resides in one database, quality performance in another database and pricing data in yet another database. Only occasionally do the individuals responsible for procurement receive access to all three databases simultaneously.
- They report that something is amiss. However, they rarely identify why something is occurring or what action should occur next.
- They lack scalability. While it may be possible to monitor the performance of twenty strategically significant suppliers by hand, it would be impractical to attempt to accomplish this task with 800 suppliers.
Supplier performance monitoring using advanced analytics overcomes all five limitations by automatically collecting data from source systems, connecting separate databases together and providing analytical tools to diagnose or predict issues with suppliers.
| Limitation | Traditional scorecard | Advanced analytics |
|---|---|---|
| Speed | Manually entered data; reflects the previous quarter | Data collected automatically from source systems |
| Trends | A 92% on-time rate can hide one item that is always late | Drill-down to part, plant and time period |
| Isolation | Delivery, quality and pricing data in separate databases | Separate databases connected in one view |
| Explanation | Reports that something is amiss, rarely why | Analytical tools to diagnose and predict issues |
| Scalability | Feasible for twenty suppliers by hand, impractical for 800 | Monitoring scales across the supply base |
Data used for supplier performance analysis
There are more systems than you would think that contain supplier-related data. An ideal starting point for collecting and analyzing this data would include:
| Data source | What it includes |
|---|---|
| Purchase order/receipt history from the enterprise resource planning system | Order date/time, requested and confirmed delivery date/times, receipt date/times, quantity and unit price received |
| Inspection results | Quality test results, rejected parts, non-conforming materials, returned products |
| Vendor invoice/payment history | Invoiced unit prices, payment terms, variations in vendor invoicing versus contracted unit prices |
| Contractual terms | Contractual unit prices, volumes, service standards, expiration dates |
| Vendor master data | Location(s) of vendors, classifications/categories of vendors, vendor name(s)/code(s), certification(s) |
| External data (if applicable) | Vendor credit ratings/worthiness, media/news coverage, country/regional/commodity-specific risk indicators |
Matching vendor records across multiple systems is typically where many projects run into trouble. Unless duplicate records are reconciled, every subsequent metric produced will be inaccurate.
Example: spend analysis conducted by an industrial manufacturing firm utilizing multiple enterprise resource planning systems was unable to produce accurate total spend analysis until vendor and material records were matched across those systems. Effort like that is unglamorous, but determines whether supplier tracking will work at all.
Important metrics
An excellent supplier scorecard measures four major areas. Each area requires a specific agreement among procurement personnel, operational personnel and financial personnel regarding the exact criteria upon which metrics will be defined.
| Area | Metric | What it captures |
|---|---|---|
| Delivery | On-time delivery | Performance against the confirmed date and/or original requested date |
| Delivery | OTIF (on time, in full) | Short shipments that on-time does not capture |
| Delivery | Lead time variability | Greater insight into planned buffer requirements than simply average lead time |
| Quality | Reject rate/defect rate | Parts per million (for high-volume components) |
| Quality | Non-conformances | Frequency and duration of non-conformances |
| Cost | Unit price variance | Variance vs contracted unit price/benchmark unit price |
| Cost | Total cost of ownership factors | Expedited freight costs, premium freight costs, rework costs |
| Cost | Commodity index trend | Trend vs current unit prices (when applicable) |
| Risk and responsiveness | Concentration risk | Concentration of spend on single-sourced vendors |
| Risk and responsiveness | Capacity/risk exposure | Geographic or capacity risks associated with suppliers |
| Risk and responsiveness | Responsive times | Timeliness in confirming orders, responding to quotes or resolving issues |
Each supplier doesn’t necessarily require each of these metrics. Strategic suppliers might be monitored on all four areas weekly. Tail-spend suppliers may only require delivery and price performance monitoring.
Levels of advanced analytics available in supplier tracking
Advanced analytics isn’t merely one tool. Rather, advanced analytics is a series of layers, and each layer answers a different type of inquiry.
Descriptive: what happened?
Automated scorecards/dashboards display current status of supplier performance continuously. Dashboards are updated directly from source systems. Descriptive analytics is the foundational layer and significantly improves upon manual spreadsheets utilized by individual users. Since all users utilize the same current data to complete tasks and make decisions, descriptive analytics represents a substantial advancement beyond manual spreadsheets.
Diagnostic: why did it happen?
Drill-down capabilities allow users to explore detailed metrics associated with a given component, e.g., part number, plant, category, buyer, time frame etc.
Diagnostic analytics is where a “supplier’s late shipments” become three specific parts shipped from one plant or a quality issue related to a new raw material provider.
Predictive: what will happen next?
Models that flag suppliers/orders at-risk prior to failure, i.e.:
- Identify which open purchase orders are likely to arrive late based upon historical behavior of supplier, size of order, lead-time and backlog
- Early detection of trends related to decreasing supplier delivery performance/quality
- Risk analytics utilizing external signals such as financial distress, news events, regional/country disruptions etc.
Prescriptive: what should be done next?
Analytic models that support decision-making scenarios/optimization. Examples include:
- Determining whether to shift volume between two suppliers
- Qualifying second sources
- Renegotiating pricing/terms
- Adjusting safety stock levels for high-risk components
Most organizations obtain the greatest value from implementing descriptive and diagnostic layers first. Once trust has been established in the data foundation (i.e., users trust the data), then predictive and prescriptive analytics can be more effectively implemented.
| Level | Question answered | Example |
|---|---|---|
| Descriptive | What happened? | Automated scorecards/dashboards updated directly from source systems |
| Diagnostic | Why did it happen? | Late shipments traced to three specific parts shipped from one plant |
| Predictive | What will happen next? | Open purchase orders flagged as likely to arrive late |
| Prescriptive | What should be done next? | Shifting volume between two suppliers or adjusting safety stock |
Supply base segmentation
Monitoring every supplier identically wastes resources. Segmentation determines where attention should be directed.
A common approach is plotting suppliers along two dimensions: amount of spend/business impact and risk/ease of replacement. This yields groups including:
| Segment | Spend/business impact | Risk | Ease of replacement | Monitoring approach |
|---|---|---|---|---|
| Strategic suppliers | High | High | Hard to replace | Weekly tracking; joint improvement plans; early warning signs of potential risks |
| Leverage suppliers | High | Low | Easy to replace | Price variances against benchmark competitors |
| Bottleneck suppliers | Low | High | Hard to replace | Ongoing monitoring for capacity/risk exposures; backup sourcing strategies |
| Routine suppliers | Low | Low | Easy to replace | Automated alerts for exceptional performance issues |
Analytics enables dynamic segmentation. As spend and performance/risk data change, suppliers move between segments automatically; no longer must wait for annual reviews.
Putting it all together in a procurement control tower
A procurement control tower provides a single view of all supplier performance metrics and outstanding orders/risks/actions taken. The layers described above are integrated into a single application used by commodity managers/category leaders on an ongoing basis.
Some organizations experience tremendous success from integrating all aspects of procurement activity into a single application. In one Kepler Advanced Analytics engagement for one of the leading commercial truck manufacturers in North America, after establishing a procurement control tower that consolidated commodity/supplier/purchasing data into a single view, lead time was reduced by 80%.
A typical procurement control tower includes:
- Up-to-date views of supplier scorecards with drill-down capabilities to part/order level
- Exception lists for late/shipment at-risk/short-shipped orders
- Price variance/contract compliance monitoring
- Alerts for concentrated spending/supplier risks/media/news coverage
- Tracking actions taken with vendors: ensure follow-up activities occur after issues are raised with vendor representatives
Utilizing spend analytics for finding larger opportunities
Supplier tracking is not only about identifying failures in supplier performance. Similar data can be used to identify cost savings opportunities.
When companies compare what each plant/business unit pays a supplier for similar products, they often discover price discrepancies. Purchasing synergies exist across business units when purchasing from the same supplier.
An example of purchasing synergy occurred when two large automobile parts producers studied purchasing practices across their respective global operations, one European-based and one North American-based. Kepler Advanced Analytics analyzed purchasing across both companies at category/subcategory level. The study included 11 categories and 121 subcategories. The study identified $56M in cost savings opportunities from the top 10 suppliers combined.
| Engagement | Scope | Result |
|---|---|---|
| Procurement control tower, leading North American commercial truck manufacturer | Commodity, supplier and purchasing data in a single view | Lead time reduced by 80% |
| Purchasing synergy study, two large automobile parts producers (Europe and North America) | 11 categories and 121 subcategories | $56M in cost savings opportunities from the top 10 suppliers |
Additional opportunities exist when performing part-level direct material cost modeling for automobile manufacturers using SAP data for tens of thousands of part numbers. Fact-based negotiating positions are created for procurement representatives in lieu of relying solely on quotes provided by the suppliers themselves.
Connecting supplier performance data to inventory management planning
Suppliers’ lead time variability impacts how much safety stock planners must maintain for their parts. As long as suppliers’ lead time variability is being monitored continually, inventory policies can be adjusted accordingly to reflect changes in suppliers’ lead time variability.
The connection between supplier performance data and inventory planning occurs both ways. Suppliers that demonstrate increased reliability justify reduced safety stocks, thereby freeing working capital. Conversely, suppliers that demonstrate decreased reliability should prompt increases in safety stocks or establish secondary sources prior to production shortages impacting production.
Similar logic exists for production planning purposes. If models indicate that several outstanding purchase orders for one assembly are at risk, planners can modify the production sequence or bring forward other production orders while the issues are resolved. Otherwise, the first indication of production difficulty is often idling equipment.
Therefore, advanced analytics for measuring supplier performance establishes its position within the broader supply chain by enabling procurement/planning/operations staff members to develop one shared perspective regarding suppliers’ reliability.
Improved supplier review processes
Quarterly business reviews improve significantly once reliable and agreed upon data exists among participants.
As opposed to reviewing each participant’s numbers for accuracy, participants begin reviewing data contained within agreed upon scorecards that each participant has previously reviewed.
Review meetings subsequently focus on fewer items than traditionally experienced, including:
- Areas where improvements in supplier performance existed
- Areas where decreases in supplier performance existed
- Specific components/plants contributing to decreases in supplier performance
- Agreed upon actions with owners/dates associated with those actions
Additionally, suppliers may analyze their own historical trends between reviews, thereby encouraging corrective actions prior to review meetings.
Joint improvement work
Shared visibility regarding forecast changes/order patterns/product defects provide significant reductions in communication required to solve problem situations among strategic suppliers.
Step-by-step guide to building advanced supplier analytics
- Define scope. Begin with one category/vendor group/region where supplier performance issues exist/were identified and accessible data exists.
- Agree definitions. Determine upon which date(s) should on-time deliveries be measured? Determine how product quality will be measured (e.g., upon receipt or when utilized)? Obtain agreement from procurement personnel/ops/finance.
- Build data foundation. Extract/join ERP/vendor quality/vendor invoice data. Clean/merge/match vendor master records across multiple systems.
- Automate scorecard. Replace manual spreadsheets with a dashboard refreshing from source systems.
- Add diagnostic views. Drill-down capability with part/assembly/plant/category/buyer/time-period combinations.
- Add predictive layer. Identify suppliers/orders at risk before failure (e.g., late order risk scoring/trend alerting on strategic/bottleneck suppliers).
- Embed routine use. Integrate dashboards into weekly supplier calls/quarterly meetings and track action items with vendors.
- Expand to additional categories/vendors once first implementation is working and trusted.
Pitfalls to watch for
- Starting with tools instead of definitions. A dashboard built on disputed metrics will be ignored.
- Underestimating master data work. Duplicate and inconsistent supplier records undermine every metric.
- Measuring everything. Too many KPIs dilute focus. Pick a handful per segment.
- Not sharing with suppliers. Performance data shared openly with suppliers often drives faster improvement than data kept internal.
Leaving out operations. Procurement owns the supplier relationship, but operations feels the impact of late or defective parts. Both need to use the same view.