Spend Data Classification -- A Pre-requisite to Spend Analysis for Strategic Sourcing By: Babhui Lee 3 rd Nov 2007
ONTENT 2 Introduction The Need for Spend Data Classification The Hurdles of Spend Data Classification Alternative Solutions of Spend Data Classification Standardization of Spend Data Classification Implementation of Spend Data Classification Conclusions & Recommendations Question & Answer
xecutive Summary Key Takeaways Spend data classification results in better spend management, and is a pre-requisite for supply management and business success. Sufficient, accurate, and timely insight into corporate spending information is vital to the success in cost reduction. Leading spend data management initiatives rely on access to all spend data sources; a common classification schema; category expertise; efficient and repeatable data cleansing and classification capabilities; advanced reporting and decision support tools; and sufficient resources and executive support. 3
The Need For Spend Data Classification
5 an You Answer These Questions? How much do we spend? On which products? From which suppliers? inadequate spending analysis capabilities are costing businesses $260 billion in missed savings opportunities annually. Aberdeen Group, The Spending Analysis Benchmark Report: Dissecting a Corporate Epidemic, January 2003
Spend Analysis & Strategic Sourcing 6 Capture, validate corporate-wide spend & suppliers Segment data Understand data (who, what, when, where, how, why) Analyze market Identify opportunities Create governance framework Align with mission and objectives Assess risk Support customer needs Leverage opportunities Identify metrics Determine quality of competition Implement process, policy & cultural transformation Develop communication plan & training requirements Measure & report performance Leverage supplier relationships Refine demand planning Communicate results Adjust & Re-initiate Spend Analysis is the first Step to Successful Strategic Sourcing
Spend Analysis 10 Best Practices Audit existing spend capabilities Classify spending at a detailed level Access all spend data sources Enhance core spend data from within and without the with vital business enterprise intelligence Adopt a common classification schema enterprise-wide Establish efficient and repeatable data cleansing and classification capabilities through the use of software or services Augment category expertise to ensure data and classification accuracy and validation Increase frequency and coverage of spending analysis Utilize advanced reporting and decision support tools Continuously expand uses and scope of a spend data management program Best Practices in Spend Analysis, Aberdeen Group, Sep 2004 7
mpacts of Inaccurate Spend Data Classification Sourcing and Supplier Management Lost Leverage on volume buy Missed savings from fragmented buy strategy Supplier Proliferation Inaccurate Spend Data Classification Poor Spend Management Compliance Reduced use of preferred suppliers with contractual terms Difficult to enforce supplier compliance to pricing, rebates and volume discount Reduced accuracy of financial reporting Inventory Management Excess stock Redundant Orders Inventory Depreciation Increased interest payme Reduced Cash Flow Product Management Part Proliferation Limited Part Re-use Design, Sourcing and Manufacturing disconnected 8
enefits of Better Spend Management Performance Area Before After Incremental Gain 1 Spend under management 42% 62% 34% Ave. Savings from strategic sourcing 6.7% 11.7% 75% Contract compliance rates 44% 59% 33% Sourcing cycle time 55 days 43 days 22% Source: Aberdeen Group, Aug 2007 1 Average incremental gains achieved by 750+ survey respondents 9
The Hurdles Of Spend Data Classification
orporate Spend Data Management Initiatives 11 BUSINESS MANDATES Implement Category Management Increase compliance CPO Purchasing Enterprise spend visibility and control Increase eprocurement adoption Manage Change Set up best practices CIO IT Deliver ROI from existing systems Improve Master/Spend Data quality across Multiple systems Ensure adoption and user satisfaction HURDLES Spend Data Quality IMPLICATIONS Restricted ROI from Spend Management initiatives
ramework for Spend Data Management We are facing big hurdles in these areas PLM Data Warehouse ERP Additional Manpower is needed to Validate and Cleanse. Whose Responsibility? 12
Why Poor Quality Data? 13 Disparate Data sources Manual Classification processes Limited Data enrichment capabilities Data quality varies within each system Data in multiple languages Manual code assignment in the source systems lead to inconsistencies and inaccuracies Faulty mapping processes Reliance on supplier data enrichment Restricted spend visibility due to inaccurate, inconsistent and non-granular classification
naccurate Classification: Business Implications 14 Commodity Code 26000 26000 26000 26000 26000 Commodity Code Description Display, LCD Display, LCD Display, LCD Display, LCD Display, LCD Item Description LCD PANEL 31.5,AU SKD,VV3A LCD PANEL 31.5,AU SKD, V2,VV3A LCD CONTROLLER/DRIVER, BGA225. DOT MATRIX LCD CONTROLLER/DRIV COMMON ROW LCD DRIVER, PQFP80. million $$$ IC Driver spend classified under LCD Display Erroneous view of Spend Reduces the negotiable spend with IC Suppliers
pend Visibility - Where Are We Now? 15 We Are Here
Alternative Solutions For Spend Data Classification
etter Data: Different Approaches 17 Spend Data Classification In-house Semi Manual Supplier Content Services Software Tool Consulting Services Hosted Solutions
ow Automation Is the Best Choice? HIGH Software Tool Hosted Solutions On- Demand Visibility Supplier Content Services In-house Semi Manual Consulting Services LOW Leveraging existing Infrastructure HIGH 18
ow Automation Is the Best Choice? HIGH Software Tool Hosted Solutions Consulting Services Detailed Visibility In-house Semi Manual Supplier Content Services LOW Repeatability HIGH 19
ow Automation Is the Best Choice? HIGH Total Cost Of Ownership (TCO) Hosted Solutions Supplier Content Services Consulting Services In-house Semi Manual Software Tool LOW Time to Value HIGH 20
utomated Spend Data Classification 21 Automated Classification Accelerated bulk Classification rate150-300k/hr Real Time classification <2 seconds Repeatable & on-demand classification User friendly Rapid implementation Ease of use Self Learning mechanism keeps operational costs very low Scalable Taxonomy independent Language & Domain independent Independent of description quality Seamless integration capabilities Powerful scheduler features Exposed Java API s Integrates with eprocurement tools and DW
pend Analysis Automation Advantage 22 Performance Area Full Automation Partial Automation Manual Spend under management 78% 66% 51% Ave. Savings from strategic sourcing 13% 12% 10% Contract compliance rates 69% 63% 46% Source: Aberdeen Group, Aug 2007
utomated Classification Is Inevitable 23 For accurate and detailed visibility over a period of time. For On-Demand Spend Analysis leveraging existing IT infrastructure. For faster time-to-value For implementation of Supply Management best practices.
Standardization of Spend Data Classification
hat is UNSPSC? 25 United Nations Standard Products and Services Code An open standard A taxonomy of products and services A practical business tool
NSPSC Structure 26 ach Level contains a 2-character numerical value and a textual description as follows Segment The logical aggregation of families for analytical purpose Family A commonly recognized group of inter-related commodity categories Class A group of commodities having a common group of function Commodity A group of substitutable products and services
enefits of UNSPSC 27 Data Synchronization across company divisions, suppliers, & global locations Process Flow Integration from RFXs, to ordering, to accounts payable, to general ledger Part Data Synchronization from design to manufacture to procurement & other systems Standard Coding System for Products & Services
alues For Enterprise 28 Automate the gathering and analyzing of spend data Provide a uniform, enterprise-wide view of spend Roll up analysis identifies contractible groups, opportunities for strategic vendor relationships Centralize procurement function, leverage volume for better pricing Collaborate with Customers or Suppliers through use of a common classification system Control maverick spend: reduce off-contract spend at higher prices Reduce inventory through product standardization
alues For Suppliers 29 Facilitate sales function, particularly through Internet exchanges Qualify as preferred supplier to customers with e-procurement initiatives Speed up new product introductions using Web services, XML, etc. Facilitate globalization of your business Collect consistent sales data across channels, regions Collaborate with customers to improve contract compliance, increasing the supplier s market share a win-win situation
Implementation of Spend Data Classification
-Step Spend Analysis Implementation Methodology 31 1. Identify Your Business Needs 2. Determine the Corresponding Visibility Needs 3. Determine the Appropriate Solution Elements 4. Determine the Appropriate Delivery Method 5. Gain Internal Support (ROI Case) 6. Evaluate Alternative Solutions 7. Select and Implement a Solution
tep 1 Identify Business Needs 32 Immediate Business Needs Future Business Needs Quantifying of savings potential Identification and eradication of Maverick spend Financial Reporting Accurately shows where funds are going and for what Identification of sources of potential savings Reduction of supply base Enforcement of supplier compliance to pricing, rebates and volume discount Reduction of Excess and Obsolete Inventory Promote part re-use instead of creating new part Show accurate spend by supplier across different naming conventions (Supplier Rationalization) Roll up spend by ultimate supplier parent Tracking of procurement process bottlenecks Provide consolidated view of new company (M&A) spend by supplier and commodity
tep 2 Determine Corresponding Visibility Needs 33 Immediate Spend Management Needs 1. Item Data Validation Completeness and accurate data 2. Item Data Cleansing Eliminate errors and discrepancies 3. Item Classification Auto and Manual methodology Future Spend Management Needs 1. Supplier Master Cleansing Normalization of Supplier Name 2. Enriched Supplier Visibility Financial Information, Credit Ratings
tep 2 Determine Corresponding Visibility Needs 34 The true picture: Data is the culprit ACTUAL SCENARIO PLM ERP E T L Garbage In Inconsistent Inaccurate Non-granular data Analytics DW Historic Spend Data DW = Data Warehouse Spend Reports Garbage out Faulty Reports Low reliability Lower ROI ETL stands for extract, transform, and load. ETL is a software that enables businesses to consolidate their disparate data while moving it from place to place, and it doesn't really matter that that data is in different forms or formats. The data can come from any source. ETL is powerful enough to handle such data disparities.
tep 3 Determine Appropriate Solution Elements 35 Complete Solution Architecture IMPLEMENTING BEST PRACTICES PLM ERP E T L Garbage In Inconsistent Inaccurate Non-granular data Quality In Consistent Accurate Granular data Analytics DW Historic Spend Data SPEND REPORTS Quality Out Accurate Reports Increased reliability Higher ROI Normalization Classification Supplier Enrichment
tep 4 Determine Appropriate Solution Delivery Method Approach Advantages Disadvantages Company Characteristics Appropriate for Manual Lower initial cost Limited Budget Approval Inconsistent Classification Limited Enrichment Not repeatable Does not scale Limited spend to be analyzed Restricted budget Short-term outlook Managed Service Quick ROI Minimal Internal Resources needed Minimal Internal commitment Best practice imbedded in vendor process Continual subscription fee Limited customization Limited integration with other applications Data security issue Poor IT-Procurement collaboration Desire to focus on core competency Want to implement an effective process, not customized to fit existing process Self - Service Max insight into own spending Min reliance on 3rd party Data security Max ability to customize Max integration with other spend management application Slowest approach Requires significant up-front organizational commitment Requires permanently dedicated resources Close IT-Procurement relationship Want to own data Want to customize solution to existing process Our Approach: Reason: Concern: We would like to take the hybrid approach. First, manual approach with self-service approach phasing in at the later stage. Need Quick Result, Plan in integrating with future application (SAP), want to own data with minimal disclosure to external parties especially our supply base. This approach requires permanently dedicated resources though resource could be reduced with self-service phasing in. 36
tep 5 Gain Internal Support (ROI - Cost) 37 Cost Element Software Licenses Software Implementation Software Integration Software Training Data Enrichment Services Total Cost of Ownership (USD, K) Yr 1 Yr 2 Yr 3 Yr 4 + Additional Hires Hardware Investment Consultation Fee Total
tep 5 Gain Internal Support (ROI - Benefits) Year 1 Year 2 Year 3+ % of Company Adopting Solution 80% 85% 100% Basis of Calculation : Company Yearly Spend = $14 Billion Benefit Type Percent Savings Affected Spend Yr 1 Savings (USD, K) Yr 2 Yr 3 Yr 4+ Increased Sourcing Savings thru effective negotiation 3% 50% 168 178.5 210 210 Improve Contract Compliance 1% 80% 90 96 112 112 Improved Purchase Efficiency 1% 90% 100 107 126 126 Improved Supplier Consolidation and Performance 1% 80% 90 95 112 112 Improved Parts deduplication & Inventory Level 0.5% 60% 34 36 42 42 Total 482 512 602 602 Sample Calculation: Savings = Company Spend x Percent Savings x Affected Spend x Adoption rate Benefit/Cost Ratio = (Total Benefits)/ (Total Cost of Ownership) N/B: Since benefits increase over time while cost decreases, the returns are greater in future years 38
tep 6 Evaluate Alternative Solutions 39 Short-list Vendors for Spend Data Management Software Evaluation Vendor 1. Ariba 2. Emptoris 3. Frictionless 4. i2 5. Ketera 6. Vertical Net 7. Austin Tetra 8. D & B 9. Analytics 10. Zycus Request for Information DONE DONE DONE DONE DONE Request for Survey Request for Demo Scheduled in Nov Request for sample Evaluation Submitted for Evaluation Submitted for Evaluation Submitted for Evaluation Submitted for Evaluation Request for Quotation
tep 6 Evaluate Alternative Solutions Survey Questions Development Data Enrichment - Service Levels (Only applicable if service delivery) a) Service Level Terms (Only apply for service delivery): 1. What % of spend do you guarantee classified? 2. What % accuracy is associated with the above term? 3. Do you guarantee a minimum % of spend classified per source system? If so, what %? 4. Are errors identified corrected retroactively or only to future data? 5. Please describe any other service level terms you offer. b) Briefly describe your data enrichment service process, including all phases. c) Describe the integration of all elements in your data enrichment service process. d) Provide a project timeline, indicating all key milestones and dependencies (both external and internal). What risk factors may affect scope? e) Describe the process for error-correction (e.g. correcting classifications). 40
tep 6 Evaluate Alternative Solutions 41 Survey Questions Development Data Enrichment - Data Normalization / Classification a) Does your solution offer automated data classification? b) Which of the following types of data does your automated solution consider for classification purposes? 1. Supplier Information: 2. Customer-specific codes (i.e. General Ledger, Material Codes, etc.): 3. Item Descriptions: 4. Other (Please Specify): c) If your solution uses multiple types of data in making classifications, are the different types of data used in parallel (all available information considered before providing a final classification) or in series (classifications based on one type of data with other types only considered when no match found)? d) What commodity taxonomies can your solution map data to? Can custom customer structures be used?
tep 6 Evaluate Alternative Solutions Survey Questions Development Data Enrichment - Data Normalization / Classification (Continued) e) How does your solution ensure consistency of classification of similar items across data systems and over time? f) Can you effectively classify direct materials spend? If so, how? g) Do you utilize a supplier database for classification purposes? If so: 1) How many unique suppliers are in it? 2) What is the geographic distribution of suppliers? 3) Is the database integrated into your product? 4) If a customer's supplier is not in your database, is it researched and added for future classification use? h) How are item descriptions used in classification? i) Describe the feedback process for customer-requested changes to classifications. j) Describe how quality assurance (QA) testing on classifications is conducted with your solution. 42
tep 6 Evaluate Alternative Solutions 43 Survey Questions Development Data Enrichment - Supplier Enrichment a) Does your solution offer supplier enrichment capabilities? b) How many suppliers do you maintain in your database? c) What is the geographic distribution of the suppliers in your database? d) What sources feed your supplier database? e) Indicate if the following types of enrichment are provided for suppliers: 1. Parent / Child relationships: 2. Credit ratings: 3. Revenues: 4. Other types of enrichment (Please list):
tep 7 Select & Implement Solution 44 Action Plan Pre Kick-off Kick Off Requirement Scooping User Acceptance Test Software Roll out Hardware Sizing Estimate no. of Users Hardware Budget Application Environment: DEV, TEST, PROD Level of integration Project Goal Implementation Plan Roles and Responsibilities Data to be collected Data customization Application configuration User reporting requirement Training Test taking by power users and core team members Desktop Application Installation Users roles and permission Go Live!!!!
Conclusion & Recommendations
onclusions 46 Spend data classification is key to successful spend management to realize cost reductions. Spend data quality or integrity is a key challenge in most enterprises. Data validation & cleansing can be timeconsuming. Effective & efficient data classification (through automated artificial intelligence) can result in detailed and repeated enterprise spend visibility. Standardization of spend data classification such as UNSPSC can reap significant benefits to enterprise. Data enrichment capability is a critical criterion in evaluating any spend analysis solution
ecommendations 47 Standardize and automate processes by adopting a standard classification schema & auto-classification. Gain executive support and build the business case. Use spend analysis to improve contract compliance. Leverage spend analysis & improved visibility to develop strategic sourcing plan. Enhance category and sourcing expertise to take action on the spend data.