Managing with blinders on: From outdated spend cubes to intelligent insights

Enterprise leaders, your focus on historical spend metrics is insufficient. The need for sophisticated intelligence solutions has never been more pressing. Historical adventures in spend classification and reporting have resulted in a realization that machine learning’s limitations are real (when it comes to invoices, “garbage in, garbage out” is true) and that obsessing over historical spend metrics is insufficient for today’s leaders. Today’s leaders must shift from simply focusing on spend analysis to advanced intelligence solutions that bring together contract, supplier, financial planning, and sustainability data. Here’s why spend cubes are failing you and how spend intelligence can transform your operations.

Why spend cubes fail modern enterprises

The term “spend cube” dates back to the late 1990s and describes the use of three dimensions to map vendor spending: Who spends the money? What do they spend money on? And what vendors get the money? To accomplish this, spend data on those three dimensions is classified into numerous hierarchical levels: internal business units and functions, commodity categories, and parent-child vendor relationships.

We have since learned that spend data is notoriously challenging to classify. Vendor invoice data is trash, and purchase orders are typically limited to the categories specified in a company’s chart of accounts. As a result, a $2,000,000 invoice from Accenture for “Cisco Services” spent against a cost center entitled “Vendor Services” using the capital expense general ledger account “623401” doesn’t help financial data sleuths. Machine learning doesn’t know if “Cisco Services” is a monthly software subscription, a consulting project, an outsourced program, or hardware maintenance. It may glean some information from prior invoice classification exercises, but that’s no guarantee. It surely won’t tell you the hourly rates of Jill Smith or the number of hours John Nguyen billed. And all of this assumes you have a purchase order to run with. Classifying non-PO spend is a task more similar to archeology than financial analysis, full of assumptions and manual error. There isn’t a finance or procurement leader alive who can state with full confidence that their spend cube is 100% accurate. It is “directionally accurate,” at best.

The limitations of historical spend metrics

The laser focus on spend in a modern world where other information is needed is a spend cube’s Achilles heel. Spend cubes are purely historical viewpoints and can’t tell you what upcoming projects are being funded. Spend cubes can’t tell you when a contract will expire and require a 60-day notice to avoid auto-renewal at a 5% increase. They can’t predict the cost of software licenses in two years based on changes in user counts and COLA changes. Spend cubes can’t tell you if a price is fair, the result of a strategic sourcing event, or if the supplier is going bankrupt. They don’t provide the nature of internal and external relationships, and if the CEO handed a $20M project to McKinsey with an ROI based on planned employee cuts. Spend cubes rarely show that Janine Johnson shipped a letter via overnight Saturday delivery UPS from Atlanta to New York with a signature requirement at the cost of $23.11. In short, spend cubes have enough data to make decisions. FP&A budgeting processes, which are almost never integrated into spend cubes, rarely are performed at a level of detail to predict true spending. Chances are that your company’s projects were presented in PowerPoint, an unstructured, high-level “data set” with supporting spreadsheets that won’t fit into your old cube. Spend cubes won’t manage your sustainability metrics required under the law. Simply put, spend cubes are outdated and lack the sophistication today’s leaders require.

Introducing Spend Intelligence—the future of spend management

It’s time for the industry to ditch antiquated spend cubes and invest in spend intelligence. What is spend intelligence? Spend intelligence goes beyond traditional spend cubes that focus on cross-tabs, pivots, and filters by leveraging GenAI technology to analyze unstructured and structured data.

Exhibit 1: Spend intelligence data sources

Source: HFS Research, 2024

The value of spend intelligence and its superiority over traditional spend cubes

Spend intelligence surpasses the limitations of traditional spend cubes. Here’s why:

1. Enhanced data integration and analysis: Spend intelligence integrates both structured and unstructured data in one place, leveraging various sources such as invoices, purchase orders, contracts, and expense reports. Unlike spend cubes, which rely heavily on predefined classifications, spend intelligence leverages advanced technologies such as GenAI to analyze diverse data types, providing a more comprehensive and accurate view of spending patterns.
2. Real-time insights: Traditional spend cubes offer a historical perspective on spending, often lagging behind current trends and needs by months while data is uploaded and manually reviewed for quality. Spend intelligence provides real-time insights, enabling organizations to make timely and informed decisions. This real-time capability is crucial for managing dynamic business environments and responding swiftly to market changes.
3. Predictive capabilities: Spend intelligence employs predictive analytics to forecast future spending trends, identify potential cost-saving opportunities, and anticipate financial risks due to both historical data and greater visibility on upcoming business plans available from FP&A budget and project presentations. Traditional spend cubes cannot ingest these data types, lack these predictive capabilities, and focus instead on past expenditures without providing foresight into future financial scenarios.
4. Contextual understanding: Spend intelligence goes beyond surface-level data by offering context about the nature of transactions, vendor performance and risk, and the fit with a company’s business plans. It can reveal details such as contract terms, supplier performance, and compliance metrics. Traditional spend cubes can’t achieve this depth of understanding, as they are limited to categorical data without contextual nuances.

Recommendations for buyers and the spend analysis industry

The current challenge is that all commercially available spend analysis tools fall short of this value proposition. We spoke to several procurement executives who are exasperated by the market’s narrow offerings. Spend analysis vendors continue to focus on spending data to get deeper category insights and improve the efficiency of reporting and data classification but lack the vision of the promise of spend intelligence. One executive for a large global retailer said, “We can’t find a solution that meets our needs, so we’re currently scoping capabilities to build our own solution.” A CPO for an airline said his organization had no spend analysis tools, but “the capabilities of the tools seem outdated in a ChatGPT world.”

Given the dearth of market offerings, HFS offers the following recommendations to buyers and spend analysis technology providers:

1. Get your data together and make it accessible: The starting point for any data-focused exercise is identifying the data you have and its sources. Because your data sources cannot be combined using traditional techniques, ignore the limitations and focus on what data you need and where you can pull it from.
2. Invest in AI and GenAI technologies: Integrate tools that can handle unstructured data. The approach to spend intelligence doesn’t rely on pulling data into common data sets. Instead, pull it into data lakes that your firm’s GPTs and LLMs can access. Then, focus on training those GPTs and LLMs to delve into insights the data can provide. Note that spend intelligence datasets are large, and IT architecture design is crucial to make this data available.
   
3. Leverage predictive analytics: Resist the initial instinct to replicate historical spend analysis. Instead, use the data to ask profound questions that proactively help in budgeting, financial planning, and risk management, including:
   1. What suppliers will customers likely need to deliver critical business projects?
   2. Predict the impact of inflation, contract terms, and corporate growth and savings initiatives on your company’s COGS and G&A ratio.
   3. Identify new suppliers offering novel approaches in your different categories.
   4. What initiatives are necessary to manage suppliers’ emerging risks?
4. Train your teams on prompt engineering: Procurement and finance teams must develop the skills to derive data from LLMs using prompts. This is a new field to these teams, and they need to be taught to develop prompts and identify hallucinations in spend intelligence recommendations.

The Bottom Line: Spend intelligence adds business context to complex business data.

To thrive in today’s dynamic business environment, enterprises must transition from outdated spend cubes to advanced spend intelligence. This shift enhances financial efficiency and provides strategic insights critical for decision-making. Enterprise leaders, the time to act is now: audit your tools, invest in AI, and train your teams to leverage the full potential of spend intelligence. Spend intelligence is not just an upgrade; it’s a necessity.