Spend Analytics: Part 3 – Applications of Spend Analysis

Parts, Product, Finance, Performance, and Cross-Functional Uses


Spend analytics can go way beyond just analyzing spend or managing suppliers. We look at some of the more novel uses of these systems.


In Part 2, we looked at how spend analytic systems can be used to manage spend, sourcing, and supplier performance and risks. Here we look at the use of these tools for parts and product management, as well as for more cross-functional uses.

Parts and Product Management

Parts Management / Rationalization

Image by Wilfried Pohnke from Pixabay

This is an example of broader uses for the spend analytics platform beyond the procurement department — parts rationalization is a cross-functional challenge where the engineering team is central.Analytics systems can unify parts data from different sourcing, engineering, and ERP systems, so that you can truly understand what you are buying and where there is unnecessary diversity of parts and materials — whether it’s steel, or semiconductors, or packaging materials. This is especially useful for companies with many divisions or to integrate new business units after mergers and acquisitions, in order to find duplication and redundancy in the parts databases. A heavy manufacturing company might have 20 different manufacturing locations all buying the same pads, and bearings, and screws, but with no master parts list, so each has a different part number and description. The commodities classification engines found in some spend analytics systems can classify the parts and automate matching and searching for similarities.

Some systems have deeper parametric capabilities, and can extract parametric data such as dimensions, color, materials, strength, etc. This can be very helpful for parts rationalization when combined with the ability to do ‘like’ or ‘near’ searching — i.e. searching within a threshold or range of values, or finding colors ‘like’ the color blue.

Visual tools can help a lot here. A graph might show all the bearings in a specific category mapped out on a bubble chart — one axis being the inside diameter, another being the outside diameter, and the bubble size being the amount of spend. This way opportunities jump out at you, seeing clusters of large bubbles near each other.


Image by Engin Akyurt from Pixabay

In Design-for-X, engineers incorporate design criteria from across the enterprise — such as ‘Design-for-service’ to make the product easier to repair and more reliable, or designing to make the product easier to ship, using common parts to make inventory management easier. Here the procurement team can enable the engineering team to reduce cost in the design phase, while also helping the supply chain team to reduce the number of parts it needs to keep in inventory and increase pooling efficiencies. The design team may be able to view failure rates and causes, based on the part, and use that to improve reliability. The system may also be able to pull in parts lifecycle data to show which parts are released, on hold, or near EOL (end-of-life).

Product Search

The same tools used to cleanse and classify data for spend analytics can be used to help people within and outside the enterprise to find products. Some systems allow their classification services to be embedded in other applications, such as a corporate procure-to-pay system or an e-commerce system, to simplify the process of finding parts or items in a catalog. Some support parametric searches by analyzing text strings in item descriptions and pulling out attributes — like for a chair it could be the size, color, and adjustability, or for a computer system it might classify by the weight of the system, screen resolution, size of the disk, amount of memory, and so forth. This could help the end-user find just the right product they are looking for in the company’s catalog. More sophisticated systems could help someone looking on an e-commerce site for a green dress — mapping colors like Jade, or emerald, to green.

Finance, Performance, and Cross-Functional Uses

Future Budgeting / Forecasting and Planning

Traditional spend analysis looks backward at what was already spent. The spend forecast can be equally important, especially when needs, prices, and spending patterns change. For example, if a company is in transition to a just-in-time replenishment model, it will impact how, when, and how much they will buy. Historical data, combined with sales forecast data can be used in the planning process to help predict what you will buy and when you will buy it.

Fraud Detection

Spend analytics can be used to look for patterns to discover anomalies in purchasing, spending, and supplier performance data that may indicate fraud, such as:

Image by Mediamodifier from Pixabay
  • Invoices in exact numerical sequence.
  • Vendor addresses that are the same as an employee address.
  • Outliers — departments or individuals spending much more in certain categories of spend than any of their peers (per person or per unit of expected consumption/output), like a factory that uses 10X the amount of certain supplies or an employee using 20X the amount of gas that their normal travel routine should warrant.
  • Orders to suspicious or potentially inappropriate vendors, such as jewelers, liquor suppliers, or other sellers of other items or services not appropriate for that department or individual. For example, charges for gasoline when the company’s entire fleet of vehicles uses diesel fuel.
  • Large numbers of transactions that are just under a manager’s spending authority limits.
  • Groups of transactions that don’t follow Benford’s Law1

Sales Analysis

Some companies flip it around and use spend analytics tools for sales analytics, bringing in the data to examine what they are selling and to whom, rather than what they are buying. Though it is a different set of data, the system for bringing in data from multiple sources, normalizing it, and then analyzing it can be useful here.

Buyer Performance / Procurement Performance / Benchmarking

Spend analytics systems can be used to look at the performance of buyers, individually or by groupings — how much are they buying, savings realized, sourcing cycle times. Of course it is important to try and compare apples-to-apples, recognizing the differences by commodity or type of business.

The tools can also be used to measure the procurement performance of a business. This is beyond just savings realized. It could include things like requisition approval times, percent of total spend managed, percent of spend on-contract, cost of procurement (e.g. cost per invoice, or procurement budget as a percent of sales), and other KPIs.

Image by Brockenhexe from Pixabay

Companies can compare performance against a baseline of past performance and/or benchmarking vs. other peers within an organization, or comparing different business units or divisions, or comparing against other firms. The latter (external benchmarking) will require acquiring the right data or indices to compare against. Again, care must be taken to understand the differences between businesses, or (especially in a highly diversified company) between the various divisions’ business units. With that caveat, benchmarking can be valuable to uncover who is doing a better job, figure out why, and then try to replicate that success in your business — or conversely highlight laggards and do some root cause analysis to fix the problems.

The beauty of spend analytic systems is that they can also be used not just for identifying performance issues, but also diagnosing the problem, enabling drill down on specific buyers or groups to see which suppliers, commodities they are responsible for, their trends over time, and other details to understand what is going on.

Collateral Benefits of Aggregation, Cleansing, Classification

Implementing spend analytics systems requires the integration, cleansing, normalizing, and classification of data from across the enterprise. This can have a number of side benefits not directly related to the spend analytics initiative. It shines a spotlight on weaknesses in the underlying data and can be an impetus to drive better data hygiene and practices across the firm. It also can serve to feed cleaner data back out to systems, or potentially even act as master data in some cases.

Some companies have found, for example, once they properly classify data about their inventory, that more accurate inventory data reduces false stockouts, makes inventory counts more accurate, driving better replenishment and purchasing decisions (not ordering things you already have and missing what’s really out of stock).

Tying it All Together

Combining Use Cases

We have spoken about the applications of spend analytics as standalone use cases. In reality, many companies combine these uses together to realize much more powerful capabilities. They may look at combinations of data including spend amounts, supplier performance, inventory levels, product forecasts, warranty data and failure rates, risk factors, external data about commodity prices, options for pricing risk mitigation (hedges/futures). These combination views can be very powerful.

Spend Analytics Across the Enterprise

It should be apparent that spend analytics tools are not just for procurement and sourcing personnel. They can be leveraged across the enterprise. We’ve already cited a couple of examples, such as parts rationalization and ‘Design-for-X,’ as well as parts search. Spend analytics can be integrated into end-users procurement tools to help them make smarter decisions. It might, for example, show the actual performance of vendors and break down their total cost to help the end-user to make a more educated decision, as well as give them some ammunition in negotiations. It could guide employees in making decisions while traveling.

The uses of spend analytics systems are almost endless. As companies master the simpler analysis and pick some of the low hanging fruit, they can build on those successes to go after more sophisticated and broader initiatives.

In Part Four of this series we cover the steps in the spend analysis process: goals and planning, data cleansing, the analysis, and taking improvement actions.

See also our library on Sourcing and Procurement / Supplier Relationship Management.


1 Benford’s law says that for most real-life sources of data, the first digit exhibits a non-uniform logarithmic distribution. Phony transactions may have a more even distribution of first digits. — Return to article text above

To view other articles from this issue of the brief, click here.

Scroll to Top