Rethink procurement benchmarking with big data | by Pasi Viljanen

Pasi ViljanenIt’s natural to be sceptical about procurement benchmarking. Most benchmarks available today have a human weakness - they are based on data gathered through surveys or interviews. Those high-level statistics for best-in-class performance you’ve seen at procurement events or whitepapers typically rely on data collected or analysed by hand. Until very recently, procurement benchmarking based on real spend data was practically impossible.

To meaningfully pursue benchmarking, professionals need spend data we can trust. Leading procure-ment organisations invest significant time and resources in spend analysis technology. This entire software category was developed to satisfy the analytics needs of procurement organisations looking to tap the potential of their spend data. Why shouldn’t procurement benchmarks have the same data-driven approach as spend analytics?

It’s only with the emergence of procurement big data that we’ve had enough technical resources to combine the extremely large sets of structured and unstructured data needed to harmonise and benchmark spend across peer organisations or industries. The results are nothing short of revolution-ary and promise to change completely the way we view benchmarks.

Big data has made the impossible possible
The reason you haven’t seen procurement peer benchmarks based on real spend data is that it’s a re-ally tricky challenge to solve on a technical level. Only a handful of companies have access to the tools and data for the task. The bigger scale means that big data or advance analytical tools needs to be used, including:
• Advanced data storage solutions to handle the enormous data volumes in a big data repository,
• Artificial Intelligence (AI) to manage the huge data variety,
• Machine learning (ML) and natural language processing (NLP) to help categorise transaction data to common hierarchies to ensure “apple to apple” comparisons,
• Complex statistical tools to ensure the veracity of the benchmarks once they have been col-lected and classified.

Peer benchmarks: challenges and opportunities
Useful peer benchmarks are created by anonymising and analysing the realised transaction data from numerous global, multibillion spend customers. We offer what we believe is the first such tool on the market. Through data-based benchmarks, professionals can compare their organisations to what is considered to be a median procurement transaction or an average big company doing global procure-ment, based on millions of data points collected reflecting real procurement spend data.

From the technical point of view, and simplifying it a little bit, peer benchmarks are created like a typi-cal spend analysis or visibility project with the same steps and tasks. The main difference is the fact that with big data benchmarking everything is hundred times bigger than normally:

• Instead of the tens of source systems we typically see with a spend analysis customers, there are literally thousands of different data sources with different kinds of data sets and data points that need to be aligned.
• We also utilise external and public data sources of data relevant to procurement.
• The amounts of spend data are much, much larger. With hundreds of billions in transaction da-ta here is much more to be harmonised and classified.
• To classify data, we’ve developed generic categories that accurately match spend data across different businesses across product categories, subcategories or geographic locations. This could not be done with the tools and methods designed for single-customer data sets.

After all the data has been collected, harmonised and classified to common hierarchies, the actual benchmark calculation is the easy part. The only remarkable thing here is to ensure that the data is rep-resentative enough to be published as benchmark. Data needs to come from several different sources and different customers. We also need to ensure benchmark data does not disclose any sensitive source data and cannot be reverse-engineered.

Benchmarking 61218
Figure 1 Procurement Big Data combines internal and external data assets at unprecedented scale

Identify and achieve world-class performance
Comparing spend performance to peers is, to our minds, the killer application for big data-powered procurement benchmarking. Within a decade, dashboards that compare own performance to peers will be essential for any progressive CPO or procurement organisation.

Through this, professionals will be able to get intuitive and dynamically updated benchmark reports that allow customers to identify world-class procurement performance metrics and track their own progress to targets over time. This gives them the power to identify the right goals, and demonstrate value achieved.

You don’t need to wait a quarter or year for updated survey data. As the benchmarks are calculated based on the realised data and the data processing is mostly automated, the updated benchmarks can be calculated and provided on monthly bases.

Collecting, cleansing and classifying procurement data is challenging enough when you have one or-ganisation with many ERPs or other data source systems. It’s even more challenging when you at-tempt to combine spend from different companies with their own taxonomies or methods to classify data. The power of big data analysis has unlocked his potential.

Pasi Viljanen is the head of procurement big data at Sievo -

TPRMS2019 Strip Banner Image 051118
Posted on December 06, 2018