Another book that was published late last year, and that has been sitting on the doctor‘s stack for review since about then, is Real-World Analytics by Michael Koukounas. the doctor has to admit that he was a bit hesitant to review this (and then lost it in the stack) because, as he just finished explaining to yet another individual before penning this post, Spend Analysis is not the same as Data Analysis, and that’s why so many companies without any understanding of the unique requirements of spend analysis for Sourcing and Procurement (who hire hard-core computer scientists who write trite like Spend Analysis: The Window into Strategic Sourcing (which is about the only book the doctor has ever reviewed that he has completely shredded) that, as it’s title suggests, gives you a cloudy window view that doesn’t give you the full picture (and often causes you to make the wrong assumptions about what is going on in the house).
But the doctor will have to admit that if you take this book as it is — a guide for building the foundation to do analytics (and not a guide for how to do them, which requires a completely different guidebook), it does a decent job. And the author — who is obviously an expert in data analytics in the Finance and Banking industry where a lot of effort goes into loan return models, credit risk prediction, and currency fluctuation models — really knows the core foundations for performing analytics quite well and does a great job discussing them.
As the author describes in various chapters, there can be no successful analytics, data nor spend, without:
- Good Data Access
and a Data Management Team - Talent
as analytics cannot be automated - Operational Knowledge
and, in particular, operational knowledge as it relates to the domain - Appropriate Trade-Offs Between Efficiency and Creativity
and fine-tuning to the audience - an Analytics Continuity Plan
in case something happens to top talent - the right teams …
data management, analytics development, and analytics maintenance - … and the right team sizes
since core development will usually only require a small team (because once the up front models are developed / implemented for the organization, new needs won’t be popping up every day), data management will require a team proportional to the number of data sources and their complexity, and maintenance will often require a larger team than you think as new data becomes available, new insights are required, and new reports are requested.
Moreover, at a high-level, the five-step game plan is correct:
- Define the Problem (and the end goal)
- Identify Touch-Points (where and when the analytics should be run)
- Understand the Touch Points (and the restrictions and requirements they place on the analytics)
- Select the Right Data (since garbage in means garbage out)
- Run the Analytics (and validate the results)
But when you start to descend from the 30,000 foot view, the details are vastly different in the spend analysis domain (and the author even implies this when he says that the analytic needs for engineers are vastly different than the analytic needs for financiers). But Real-World Analytics is a great guide to getting the precursor foundations right.