Home > Resource Center > White Papers > BI and Purchasing

Beauty and the Beast: BI Meets Real World Purchasing

Download the PDF

Summary:

Software for purchasing departments has tended to confine itself to automation and communication. While a few Business Intelligence applications, primarily aimed at extended supply chain management and/or reverse auctions exist, they almost always require high volumes, power users and substantial financial and change management commitments. What's brand new on the scene is operationally embedded business intelligence applications for purchasing that are lightweight, easily implemented and deliver major benefits.

Introduction

During the last decade, almost all the sophisticated software attention garnered by purchasing departments has been focused on the extended supply chain. Collaborative Supply Chain Management (SCM) tools and reverse auction applications are transforming the way high-spend, high volume corporations are doing the business of purchasing. In these tier one multinationals, Business Intelligence (BI) applications have started to make inroads, moving out from corporate strategy and statistics to start to power operational and tactical decisions in both SCM and SRM (Supplier Relationship Management). But for those purchasing departments handling hundreds, not thousands, of suppliers, and with spend in six and seven figures, not nine or ten, most of the implemented applications have tended to simply automate the way purchasing departments have always done their jobs. And, for these purchasing departments, business intelligence applications still seem anything but near term.

While the vanguard has grabbed the headlines, the majority of purchasing departments are struggling to keep pace with an ever more exacting marketplace. Current estimates are that only 50-60% of US companies use even slightly advanced SCM/SRM techniques, fewer than 1 in 10 are positioned to fully exploit the concept of strategic procurement, and less than 1 in 100 has actually fully implemented it. But, all purchasing departments are subject to the same market forces, which recently has meant a rapid increase in materials costs. Global sourcing helps to address rising costs, but often contributes to greater deviations in lead times. Increasing the supplier base also tends to increase supplier instability, as less proven partners enter into the database. The need for advanced capabilities certainly exists, but most mid-size companies are put off by high cost and the significant change management commitment required by many of the new solutions.

Purchasing Software

Purchasing departments historically existed in the backwater of the back office. Until recently, their services were critical, but certainly not sexy. They originally came into being as a separate organizational function to channel dollars through a single department dedicated to cost savings. Purchasing agents soon developed paperwork systems and forms to speed work and ensure that all transactions were legal, ethical, and properly recorded.

During the last 30 years, technology has increasingly and systematically simplified much of this work. A quick scan of the software available for purchasing shows that the great majority of applications actually implemented in purchasing departments aimed at automating existing processes, not fundamentally altering them. Software now facilitates everything from requisition to payment, including streamlining RFQs, creating POs, PO management, order acknowledgement, order processing, blanket orders, automated approvals, and creating supplier report cards.

More advanced technology aids in communication, both internally, with global demand aggregation, integrated accounting, and inventory management, and externally, via internet based procurement.

AMR Research separates the purchasing function into procurement, with goals of efficiently executing purchasing processes, and sourcing, which strives to dynamically reconfigure the supply base and manage suppliers for maximum effectiveness and profitability. Until recently, procurement software that improved purchasing processes definitely dominated the landscape.

Yet purchasing surveys routinely report that purchasing department personnel spend a greater percentage of their time on sourcing than on any other single activity. And, sourcing is becoming ever more complicated now that purchasing no longer just entails finding the lowest cost supplier, but is tasked with optimizing across a number of variables.

According to John Trush, supply chain and purchasing specialist and partner with IBM Consulting, "It has been a genuine challenge for many companies to assess suppliers not only on cost, but also on quality, service, lead time, financial viability, profile of goods procured, and technology. Yet only with this kind of broad spectrum analysis and understanding of hidden costs and benefits can companies remain competitive and agile."

One reason is that there's just no good way to do that without BI tools. Multiple and frequently interdependent variables quickly raise the computational bar above human capabilities. Factoring in time constraints and a workforce typically unskilled in complex analytics means that without BI tools, purchasing departments simply will not be able to meet these challenges.

Business Intelligence Software

Business intelligence (small "b", small "i") started out simply enough with IT departments providing standard, pre-defined reports for operations personnel and management. Reports would be issued weekly or monthly and acted as hard copy dashboards for the organization and departments within it. As time frames collapsed and the need for better, timelier information grew, so did the requests coming into IT, often creating report request backlogs. Eventually, ERP systems sprouted report writers to help individual users and departments take control of their own data requirements. But, even today, the most popular feature in any reporting system is "Export to Excel"- canned reports usually miss the mark for the needs of individual users without additional massaging. All too frequently, individual spreadsheets end up acting as central repositories for critical corporate information.

Business Intelligence (capital "B", capital "I") more often describes big picture, "insight" tools - software specifically designed to help companies understand what makes the corporate wheels turn and to predict the future impact of current decisions. BI encompasses canned analytics, independent business intelligence processes, data mining decision support, and analytic technologies. The programs search databases using techniques such as neural networks and decision trees, looking for telling correlations and patterns that are virtually impossible for humans to detect, and serve them up to help executives steer the corporate ship.

Most BI systems operate without knowledge of the underlying domains they're investigating, however, making power users critical to ensure correct interpretation. Without the fail safe of technologically savvy users, BI's ability to deliver executional support is hampered. Consequently, Gartner VP Howard Dresser, observes "While some inroads have been made to bring BI out of corporate strategic planning and into operations, BI has lived, to a great extent, in the very rarified world of statisticians and analysts."

The brand new promise is that BI tools are being formulated that will help rank and file employees harness data too complicated for manual manipulation. Some new BI business applications, including those from Cognos, Business Objects, SAS, and others are designed to support day to day operational blocking and tackling. Applications from Informatica and CombineNet specifically target purchasing.

Informatica's PowerAnalyzer analyzes purchasing data to ensure that buyers around the world are taking advantage of negotiated deals. The tool automatically alerts buyers when they exceed spending thresholds that entitle the company to discounts.

The consumer package giant P&G is using CombineNet, an electronic expressive bidding tool that relies on combinatorial science and gives suppliers the power to change economic order quantities, product bundling, delivery routes and timing or other variables to optimize efficiency.

SAS has recently introduced SAS 9, a BI heavyweight that promises to deliver "fit to task" intelligence. Data integrated from every corner of the enterprise is used to provide decision support capabilities to functional areas within the organization, including purchasing. SAS promises that this wide-ranging system will provide a "single, verifiable version of the truth."

Although the price tag on these tools varies considerably, they tend to be big ticket, big company items. In addition, Forrester analyst Keith Gile recently argued that price is not their only drawback, stating, "Today's BI vendor products lack a consistent mechanism for defining, managing, and implementing or inheriting prescribed business processes from external applications. Consequently, BI-based performance management is useful only to a handful of employees that target strategic initiatives based exclusively on the available data, and not the processes that the data supports."

For the masses, true usability will mean combining the technology with common applications, integrating BI with existing business systems and processes. BI that's as easily deployed and rolled out as a new desktop application. Bottom up BI.

Some progress appears to be emanating not from the BI establishment whose point of view is, almost by definition, external to manufacturing, but from the very software vendors whose products are already deeply involved in operations: ERP providers.

Case in Point

For the last twelve months, TriContinent Scientific, a Hitachi Chemical Company providing liquid handling products and instrument components to the medical diagnostics and biotechnology industries, has been using new software to streamline the operations of their purchasing department. The software was developed by Relevant Business Systems, a provider of Enterprise Software for project-oriented businesses, primary in Aerospace & Defense and Contract Manufacturing. TriContinent employs Relevant's ERP system throughout their organization and since last year, has used Relevant's new software, Business Wizard (BW) to reduce the time and steps required to acquire or retrieve, analyze and act on critical information. Recently, Relevant added a Predictive Analytics capability to Business Wizard, greatly enhancing a purchasing department's ability to optimize sourcing on a PO by PO basis.

Business Wizard looks like a management dashboard that accesses legacy data, ERP systems, and off line data in Access databases or data warehouses. It sits like a mantle over disparate software within an organization, is database independent, and can be used throughout an enterprise or on a single desktop. End users create consolidated, tailored views of the data their job function requires, using pre-existing templates as a basis, or by creating a totally custom view. Business Wizard assembles a "model" of what a buyer has identified as important data, pulling that information from disparate databases in the ERP or other systems and presenting it for the end user in real time. Business Wizard's self-serve, end user driven templates encapsulate the domain expertise of both the buyer and the underlying manufacturing system and provide a major point of difference from other business intelligence tools. With the newly added Predictive Analytics capability, BW provides complex analysis on the fly at the behest of the end user.

Prior to implementing Business Wizard, TriContinent followed fairly typical sourcing steps. John McDaniel, Materials Manager for TriContinent described the process as four basic steps:

Establishing "We need to buy this": Assessing the current inventory position. MRP (Materials Resource Planning) creation of purchase requisition information
- What item needs to be purchased
- What is the item's description
- When must the item be in-house and available in order to support manufacturing
- What quantity is required

Starting programs to execute Requirements Analysis
- Pull up screen showing dates and demand history for the part in question
- Execute requirements analysis

Determining how to get the best price
- Run a usage report for the item, determine if typical usage is high or low and whether or not to go to a higher order quantity to obtain a price break

Opening the Purchase Item Vendor Menu to get a look at all possible vendors, pricing, delivery times and quality
- Analyze all the information and select the vendor and order quantity
- Execute a purchase order

These steps required that purchasing employees had fairly sophisticated system navigation skills and good familiarity with where all the required information resided in the underlying software. Vendor analysis was a best judgment call, based on available history and the extent of the employee's ability to mentally balance multiple variables.

John McDaniel used Business Wizard to create a single view that showed all the information required to complete each of the four procurement steps in three separate panels, showing:

Requisition information from MRP. In this panel, highlighting an item number links to Parts Demand information

Parts Demand information, from the work order side of the ERP system. Right clicking puts the user into Requisition Maintenance, where he/she can change any information without leaving Business Wizard. Tabs take the user to
- Inventory information
- Usage for 2004
- Usage for 2003

Vendor Information. Displayed information includes
- Name
- Last 3 purchase orders and pricing
- Quality

A tab allows the user to look at any open orders for that vendor.

Within these views, Purchasing could right click and launch any underlying ERP program and effect changes in it, without closing down other programs or leaving Business Wizard.

Business Wizard showing Panel 1: Requisitions Information from MRP. Panel 2: Receipts Information with details on Days Late and Quality Performance (% RTV) highlighted. Panel 3: Vendor Information with Supplier Rankings

With Predictive Analytics

To the basic Business Wizard tool, Relevant has added a layer of sophistication that could make it Purchasing's "killer app", by enabling buyers to quickly assess how suppliers can be expected to perform against multiple variables on a given purchase order or, to predict which vendor can be expected to perform best, given the parameters of a specific purchase.

Companies primarily concerned with cost who purchase from a small number of vendors can fairly easily and accurately rank their suppliers. Add an additional metric, such as "days late" to the to the mix, and evaluating even ten vendors of a single product across hundreds of purchases begins to require spreadsheet agility. As soon as evaluation is expanded to include multiple criteria (quality, shorts, financial terms, etc.), or interacting or potentially correlated variables (lateness and shipping options, for example), accurately scoring vendors becomes very computationally complex.

Finally, the overall performance rating of a vendor may hide very good or very bad performance on certain types of orders. For example, a PCB vendor may have an excellent performance history for boards with less than 10 layers and standard delivery times, but may perform abysmally on quick turn, high layer boards. Most supplier ranking programs cannot computationally find and accommodate such subtleties.

BW's Predictive Analytics detects the patterns and finds the attributes in company's transactions history that enable finely tuned prediction of future vendor performance. It accomplishes this without separate data input and without running software isolated from the firm's underlying technology structure. It works within existing purchasing processes to enhance them.

With finely tuned analytics, purchasing personnel can easily look at anticipated vendor performance on specific order types at the moment when a purchasing agent needs the information to execute a purchase.

Predictive Analytics Analysis screen shot, showing:
Top panel: BW dataset of receipt history on a given part. In this example, variables shown include quantity, days late, rejects (RTV), price, vendor & carrier

Bottom panel: Within the dataset, PA identifies fields with predictive capabilities (determinants). Determinants are weighted for importance and the resultant formula is used to return a recommendation.

Pulling information directly from on-going operations, Business Wizard constantly feeds its predictive analytics capability real-time information. The predictive analytics feature continually relearns, tracking vendor performance on an on-going basis. As vendor strengths ebb and flow, purchasing agents are always making their decisions from performance predictions based on the latest vendor information.

Business Wizard Predictive Analytics combination delivers sophisticated analytic capability to the task of sourcing. Its direct link between operational data and the business intelligence allows non-statisticians to confidently take the reins.

Conclusion

Market demand is forcing the need for sophisticated BI tools down into the organizational depths of even mid-sized companies. One area increasingly facing this pressure is the purchasing department, where the ability to conduct multi variant analysis of vendors is becoming ever more critical as cost pressures mount. Still, with seemingly few exceptions, BI tools' inability to tap into and exploit processes as well as data hampers their ability to deliver on their operational promises. And their dependence on statistical and analytical capabilities beyond those typically found in mid-sized purchasing departments further distances their deliverables from the end users most in need.

Perhaps the market is poking at this problem from the wrong direction. Large, invasive, big stick BI tools may not end up doing the job half as well as the less ambitious but, at least at this point, more functionally adept applications rising up from the manufacturing domain itself.