Claude Code Basics for S2P

Spend Analytics

CookBook

Please ensure that all required files are downloaded before you begin working through the chapter. Having the files readily available will help you follow the content smoothly and complete the activities without interruption.

Data files for all the chapters

Data Files

Your spend is scattered across Oracle EBS, Fusion, SAP, and spreadsheets and nobody trusts the numbers. This recipe shows you how to map fields across source systems, build a unified spend mart, validate completeness, and reconcile totals using Claude Code. What traditionally takes a data engineering team weeks takes you hours.

Use Claude Code to Consolidate Spend Data from Multiple ERPs into a Single Trusted Spend Mart

Consolidate Spend Data from Multiple ERPs


"IBM," "I.B.M. Corp," and "International Business Machines" are the same supplier but your ERP does not know that. This recipe uses Claude Code's semantic understanding to identify, group, and resolve duplicate supplier names across systems, producing a canonical mapping table and a reusable skills file. Traditional analyst effort: 2 to 3 days. With Claude Code: 15 minutes.

Use Claude Code to Dedupe and Normalize Supplier Names Using Semantic Intelligence

Dedupe and Normalize Supplier Names

companion reference guide to the classification lab. Covers the full UNSPSC taxonomy hierarchy across Segment, Family, Class, and Commodity, and walks through how Claude Code handles the classification chaos when every business unit uses different category names for the same thing. Includes expert tips and further exploration for building classification accuracy over time.

Use Claude Code to Auto-Classify Thousands of PO Lines into UNSPSC Taxonomy at Scale

 Auto-Classify PO Lines into UNSPSC Taxonomy

The reference companion to the dashboard lab. Covers the data pipeline architecture, KPI framework design, threshold-based alerting logic, and the automation skills file that makes next month's refresh instant. Organizations with real-time spend dashboards report 15 to 20 percent faster decision cycles and 5 to 10 percent better budget adherence.

 Use Claude Code to Build a Near-Real-Time Spend Dashboard Without a BI Team

Near-Real-Time Spend Dashboard


80 percent of your suppliers account for less than 5 percent of spend and each one costs between $500 and $2,000 per year in administrative overhead. This hands-on lab teaches you to define tail spend thresholds, cluster similar suppliers semantically, quantify consolidation savings, and generate an actionable report for category managers. Traditional effort: 1 week. With Claude Code: 20 minutes.

Use Claude Code to Identify Tail Spend Consolidation Opportunities and Quantify the Savings

Identify Tail Spend Consolidation Opportunitie

Every PO that bypasses an approved contract is money left on the table. This recipe teaches you to intelligently match PO data to contract records, classify and categorize maverick spend by type and root cause, calculate the financial leakage, and build a skills file that automates detection on an ongoing basis. Traditional effort: 2 to 3 days. With Claude Code: 15 minutes.

Use Claude Code to Detect Maverick and Off-Contract Spend Before It Becomes a Bigger Problem

Detect Maverick and Off-Contract Spend


The monthly spend cube takes 2 to 3 days of analyst effort every single month, covering ERP extracts, data transformation, aggregate calculations, total validation, and variance explanations. This recipe automates the entire pipeline using Claude Code and encodes it in a skills file so next month's cube takes 10 minutes instead of 3 days, with built-in validation and prior-period variance analysis included.

Use Claude Code to Automate Monthly Spend Cube Generation from 3 Days to 10 Minutes

Automate Monthly Spend Cube Generation

Finance and procurement report different spend totals for the same period because they use different FX conventions. This recipe teaches you to apply the right conversion methodology for each use case, handle functional versus reporting currency correctly, decompose spend changes into volume and currency components, and reconcile your numbers with finance. Traditional effort: 1 to 2 days. With Claude Code: 10 minutes.

Use Claude Code to Normalize Multi-Currency Spend for Accurate Global Reporting

 Normalize Multi-Currency Spend for Global Reporting

A category manager celebrates a 3 percent price reduction but the market index dropped 12 percent in the same period. That is a 9-point underperformance hidden in plain sight. This hands-on lab teaches you to map categories to appropriate commodity indices, align time periods, calculate procurement versus market performance, flag outlier categories, and generate a market intelligence report ready for negotiation. Traditional effort: 1 week. With Claude Code: 25 minutes.

 Use Claude Code to Benchmark Your Category Spend Against Live Market Commodity Indices

Benchmark Spend Against Market Commodity Indices 

IT spend spiked 40 percent this month and the CPO wants to know why. Three hours later, an analyst finds the answer: one PO, one non-preferred supplier, one project manager who bypassed procurement. This recipe builds the connected data model that makes that same investigation take 5 minutes instead of 3 hours, linking spend summaries to PO details, organizational hierarchies, and contract compliance in a reusable investigation agent. Traditional effort: 2 to 3 hours per investigation. With Claude Code: 15 minutes.

 Use Claude Code to Drill Down from Any Spend Anomaly to Its Root Cause in Minutes

Enable Drill-Down from Spend Anomaly to Root Cause

<h2 id="f8ef678d-e4ec-43f8-ba89-7bbbf6146c1b" data-toc-id="f8ef678d-e4ec-43f8-ba89-7bbbf6146c1b" style="font-size: 24px; line-height: 1.2;">Spend Analytics Cookbook Using Claude Code</h2><hr><h2 id="cbbef01b-491c-492d-9826-b9dd384549c5" data-toc-id="cbbef01b-491c-492d-9826-b9dd384549c5" style="font-size: 24px; line-height: 1.2;">Short Description (247 characters)</h2><code>A practical recipe collection for procurement professionals who want real results fast. Learn to classify spend, detect leakage, benchmark against markets, and automate monthly reporting using Claude Code — one hands-on recipe at a time.</code><hr><h2 id="60482e1b-1ae4-46d0-96f7-7865b47e0000" data-toc-id="60482e1b-1ae4-46d0-96f7-7865b47e0000" style="font-size: 24px; line-height: 1.2;">Long Description</h2>Most procurement analytics training teaches you concepts. This course teaches you recipes.Spend Analytics Cookbook Using Claude Code is a hands-on, practitioner-first course built around one simple idea: every spend analytics problem your team faces has a repeatable, automatable solution — and Claude Code is the engine that makes it possible without a data science team, a BI platform, or weeks of analyst effort.This is not a theory course. There are no abstract frameworks or hypothetical scenarios. Every lesson in this cookbook is a working recipe: a real procurement problem, a step-by-step Claude Code solution, a reusable skills file you keep forever, and a clear before-and-after showing exactly how much time and effort you saved.Across 14 recipes and 10 core spend analytics use cases, you will work through the complete spend analytics challenge set that procurement teams face every day. You will consolidate spend data from multiple ERP systems into a single trusted spend mart. You will deduplicate and normalize thousands of messy supplier names using Claude Code's semantic intelligence. You will auto-classify PO lines into UNSPSC taxonomy at scale. You will build a near-real-time spend dashboard without a BI team. You will identify tail spend consolidation opportunities, detect maverick and off-contract spend, automate your monthly spend cube generation, normalize multi-currency spend for global reporting, benchmark your category prices against commodity market indices, and build a drill-down investigation agent that traces any spend anomaly to its root cause in minutes.Every recipe follows the same proven structure: the pain point that makes this problem real, the business context that makes it worth solving, a step-by-step Claude Code walkthrough, a reusable skills file that encodes your solution for future use, success metrics to validate your results, and pro tips that take your output from good to expert-level.This course is for you if:<ul><li>You are a procurement, finance, or supply chain professional who knows your spend data holds answers your team cannot currently access</li><li>Your monthly spend cube takes days to produce and is already out of date by the time it reaches leadership</li><li>You want to detect maverick spend, classify unstructured PO data, and benchmark against market indices without hiring analysts or buying expensive software</li><li>You are an Oracle ERP consultant or SI practitioner who wants to add spend analytics automation to your client delivery toolkit</li><li>You have completed foundational courses in this series and are ready to build a working spend analytics capability recipe by recipe</li></ul>What makes this course different: Most analytics courses give you knowledge. This course gives you tools. Every recipe produces a working, reusable output that you take back to your organization and deploy immediately. The skills files you build across the 14 recipes accumulate into a personal spend analytics automation library, one that gets smarter and faster every time you use it. And unlike vendor-specific training that only works inside one platform, every recipe in this cookbook works with your existing data in whatever format it lives in today.By the end of this course, you will have built a complete spend analytics automation library using Claude Code covering data consolidation, supplier normalization, spend classification, dashboard generation, tail spend analysis, maverick detection, spend cube automation, currency normalization, market benchmarking, and anomaly investigation, along with a clear implementation plan to deploy each capability in your own organization.

A practical recipe collection for procurement professionals who want real results fast. Learn to classify spend, detect leakage, benchmark against markets, and automate monthly reporting using Claude Code — one hands-on recipe at a time.

Spend Analytics Cookbook Using Claude Code

Sambit Tripathy

Level

What you will learn

Build a unified spend mart using Claude Code that consolidates data from Oracle EBS, Oracle Fusion, SAP, and spreadsheet sources into a single trusted spend foundation that every procurement stakeholder can rely on

Classify thousands of unstructured PO line descriptions into a standardized UNSPSC taxonomy using Claude Code as your classification engine, replacing weeks of manual analyst effort with a 30-minute automated recipe that gets more accurate over time

Detect maverick spend, off-contract purchasing, and tail spend consolidation opportunities using Claude Code so that policy leakage and missed savings are surfaced automatically rather than discovered accidentally months later

Automate your monthly spend cube generation, multi-currency normalization, and near-real-time dashboard refresh using Claude Code so that your team stops spending days on data preparation and starts spending that time on decisions

Benchmark your category prices against live commodity market indices using Claude Code so that every negotiation is informed by external market reality rather than internal historical baselines alone

Build a personal spend analytics automation library of reusable Claude Code skills files across all 10 recipe areas so that every problem you solve once is solved permanently for every future month, category, and data refresh that follows

Course content

Course Description

FAQ