Data Transformation

Data Transformation

Getting your spend data into the best possible shape

Spikes Cavell were the obvious choice when selecting a partner for spend analytics: Their knowledge and understanding of data as a whole is vast, and their attention to detail second to none.  Unlike many providers, they pride themselves on ensuring the data is fit for purpose before making it available for analysis. The robust line level checks they undertake mean that the data is cleansed, de-duplicated, categorized and enriched to such a high standard it gives complete confidence in the subsequent data analysis.

Rob Peck, Procurement for Housing (PfH) (UK)

We believe business intelligence should be driven by business objectives and strategic intent. And, if your intent is strategic, then the data you use to underpin your decision making needs to be as complete, accurate and reliable as is feasible. It’s all about confidence. And that confidence will evaporate rapidly if your spend data isn’t fit-for-purpose.  Of course the tools matter, but in our customers’ experience the data matters more. That’s why we spend as much time, effort and energy as we do focused on getting your data into the best possible shape.

But transforming spend data into actionable business intelligence is tough. Technology can certainly help, but in our experience, if the raw data is incomplete, lacking in detail or not coded accurately at source, then no auto-classification or artificial intelligence based approach will deliver an acceptable end result for more than a small proportion of your spend. Simply put, if a reasonably knowledgeable human can’t tell you what it was you bought by looking at the data, then neither can a machine. And rules based mapping of your account codes doesn’t add a great deal of analytical value either. If a code was wrong or inappropriate before you mapped it, it will still be wrong or inappropriate once it has been mapped to something else. We’ve lost count of the number of occasions where organizations have invested time, money and effort in auto-classification or code mapping only to end up bitterly disappointed.  That’s why we do it differently.

Our hybrid data transformation methodology is an effective combination of people, process, data and technology that has been designed to deliver the best possible analytical outcome regardless of the quality of the source data. And yes, we use sophisticated auto-classification and machine learning, but only where it’s appropriate to do so. Our key difference is our use of people. Analysts are actively involved throughout the data transformation and enrichment process, making decisions that no business rule or computer program can. That’s why we can confidently commit to correctly categorizing 97% of your spend by value.

A Solid Foundation

No-one would ever risk putting hours of work into building a house if it lacked a firm foundation. The same is true of spend analytics. Good data, that is fit-for-purpose, is the foundation on which procurement strategy and operational execution rests. That’s why we pay particular attention to ensuring that the raw data you extract from your accounts payable, purchase card and eProcurement  systems properly reflects the totality of your spend on goods and services. You may be surprised at how little you need to give us in order for us to give you a lot more back and how little effort you need to expend to give it to us. Eight to sixteen hours total effort on your part is all that is typically required to get us the data we need, and to review the results of the more than 400 checks our Resequence data validation engine routinely makes as part of our initial Data Fitness Check (DFC) process. Once the Resequence DFC is signed off there is nothing more you need to do until our transformation work is complete and your data is published to the Observatory.

I chose Spikes Cavell not only because they were able to transform our spend data into something ready for analysis, but also because the pre-built reports in the Observatory meant that I had immediate access to exactly the information I needed, giving me some firm foundations to stand upon.

Denislava Ivanova, Homes for Haringey (London, UK)

Cleanse, De-duplicate, Categorize and Enrich

We don’t believe that merely reorganizing the data that you gave us adds nearly enough value. When we say we transform, enrich and add value to your data that’s precisely what we mean. But then we do have several significant advantages. Our core internal reference database has been painstakingly constructed as a result of processing half a trillion dollars (£320bn) of spend with 9.5 million vendors. Our commodity categorization training set is derived from categorizing more than 40 million purchased items to UNSPC. The reference database we license from Experian contains extensive organizational profiles of almost 45 million US and UK businesses, government bodies and not-for-profits. And we bring all of these data assets to bear when cleansing, de-duplicating, categorizing and enriching your data.

We also throw smart and experienced people into the mix. Our project managers and analysts are actively involved throughout the data transformation process, making decisions that no business rule or computer program can. Technology has a part to play too. Over the years we’ve invested heavily in the development of a sophisticated, purpose built infrastructure to manage the process of transforming your data. That’s how we’re able to deliver enterprise-wide spend visibility across large, complex, multi-faceted organizations comprised of many disparate parts without skipping a beat, or missing a single piece of data.

We recently contracted with Spikes Cavell to conduct a spend analysis across all 23 CSU campuses and the Chancellor’s Office. It was a tough year to start this project, as we were in the middle of transitioning to a revised financial system as well as a new procurement card provider. We ended up delivering as many as four datasets per campus. Despite those challenges, Spikes Cavell was able to complete the project two weeks ahead of schedule with the team being beyond helpful. We are already zeroing in on initial opportunities for collaboration and savings – and we expect a great deal more moving forward. Spikes Cavell has performed beyond our expectations.

Tom Roberts, California State University System (California, US)

How the data is organized for analysis matters

Getting quickly to the answers you need, and discovering hidden insights in your data, is about more than the tools. How the data is organized for analysis matters too. Over the years we’ve used every taxonomy for categorizing spend available and none of them did a perfect job of balancing the need to drill to detail with summary labels that made sense. So we analyzed almost half a trillion dollars (£320bn) worth of actual expenditure and built our own.

vCode is a three level hierarchical taxonomy for categorizing spend that perfectly balances the need for detail (almost 600 categories at the lowest level) whilst enabling summation (30 categories at the top level) in terms that procurement, finance and management can comprehend. It has also been built to facilitate meaningful categorization of spend on goods and services where there’s little or no detail in the source data. If you want to generate reports and analysis using something else that’s fine too. Our analytical tools support most of the taxonomies you’ve heard of, and a few you probably haven’t.

We live and breathe data

At Spikes Cavell we love data. For almost 25 years we have been capturing, categorizing, organizing and presenting data so that it can be effectively leveraged to deliver the answers you need, quickly, reliably and with little effort on your part. Our data transformation methods, tools and processes have been specifically designed to turn poor quality data into business intelligence that is fit-for-purpose. So no matter how bad you think your raw data is we can almost certainly do something valuable with it.

If you’d like to know more about how our hybrid data transformation methodology then please Contact Us.

Spikes Cavell Data Processing Diagram

Bookmark and Share

Comments are closed.