How Do You Define and Measure Accuracy in Spend Analysis?

A lot of spend analysis vendors use percentages when describing spend analysis and their capabilities. “Classified to >85% accuracy.” “Cleansed at 95% accuracy.” Or “Classified close to 100% accuracy.”

At Sourcing Force we also use percentages to validate accuracy with our spend analysis. But what does this mean and what should you be looking for when evaluating a spend analysis?

Why is this so important?

During spend analysis, data is collected, cleansed and classified with the goal of analyzing the data to make business decisions.

But if the data isn’t cleansed and classified correctly, poor business, financial and strategy decisions will be made.

Who confirms the accuracy of the data?

When using a third-party solution provider or a consultant to conduct your spend analysis, you need to ensure that you will be the one determining the accuracy of the results, not the solution provider or the consultant.

What may be considered accurate supplier classification, or transactional classification for one customer, may not be accurate for another. 

The supplier Polaris is often used an example where the opportunity for misclassification exists. Spend for Polaris may be classified as a snow mobile, a missile, or as a heart monitor.  Another example is a laptop case.

Does your company classify laptop cases as a Computer Accessory, as Luggage and handbags and packs and cases, as Office Equipment and Accessories and Supplies, or something else? 

What method is used to determine accuracy?

Whether using a third-party solution, a consultant, or performing a spend analysis in-house, you need to use some sort of method to determine the accuracy of the analysis.

Consider it an audit, or quality control, but you need to ensure that the data you’re analyzing is accurate before you can trust it to make strategic business decisions. And you need to determine what level of accuracy is acceptable to you.

At Sourcing Force we use a formula that generates a representative sample of an overall data set that is approved by the client. Regardless of the method used, you need to identify and approve the method prior to the spend analysis.