Statistics: Lies, Pliable, Wrong?
Everyone knows “There three types of lies – lies, damm lies, and statistics” – Benjamin Disraeli,
But how about “Facts are stubborn things, but statistics are pliable” – Mark Twain,
or “All models are wrong. Some models are useful “– Chip Frank
In the era of big data and analytics it is best to keep in mind some of these sayings. Time and time again I run across companies who are looking to implement or who are implementing analytics across the company. The discussion goes something like this:
- We want to have KPI’s that go across the enterprise
- Everyone must be measured by the same KPI’s
- These are the KPI’s
- What is the definition of the KPI’s
- We can’t agree on the definition of the KPI
- We can’t find the data
- Each plant has different data sources
- Nobody agrees on anything
- Where the data, what processes do we need
- Mr. X has been looking into this for 6+ months
- What does SAP have, can you help us?
- Where do we start?
While is great to have a corporate goal of common analytics to run the company, and to some people this is not just a nice to have, but is mandatory, it is not always an easy task.
Different measurements, difference data, and different interpretation of both the data and the measurement are just some of the hurdles that need to be cleared before a common set of KPI’s can be generated.
On top of this confusion around the makeup of the KPI are the whole data collection & calculation processes themselves. One company that I know of, had a process that allowed different people to alter i.e. ”correct” the data as it progressed through the process. While this was put in place for the most innocent of reasons (not having to correct the originating data e.g. speeding up the process) it was open to manipulation. In fact I have heard comments saying that if
“management knew the truth, …, they can’t handle the truth”, truly scary. With different areas of the company having different data sources, different calculations, and different interpretations of the KPI’s how can you trust anything? Is it even possible to have a commona set of analytics? Yes of course it is, but it is not an easy task.
So where do we start? My suggestion is to start with what you already have, while you investigate your desired set of Global KPI’s. SAP comes with a variety of standard KPI’s and reports both as part of SAP ECC and Business Intelligence
In logistics in SAP ECC, for example there is the Logistics Information System which uses informative key figures to aid in detection of existing problem areas and to analyze their origin. Business transactions that are often performed thousands of times in one day form the basis for these key
See this link for more information on these information systems. http://help.sap.com/saphelp_erp60_sp/helpdata/en/c1/375492449a11d188fe0000e8322f96/frameset.htm
In addition there is also the BW content and content extension packages. BI Content is a preconfigured set of consistent metadata in SAP BW that represents the entire process flow from extraction of data originating in SAP.
The following link will give details of one such content package.
Within the contact packages there a large amount of queries and KPI’s (key figures) in each functional area that can be used to further analyze your operations. The following link shows you the queries that are delivered in the plant maintenance / enterprise asset management area http://help.sap.com/saphelp_nw70/helpdata/en/f2/7a0c3c40787431e10000000a114084/frameset.htm
See http://help.sap.com/bicontent for all BI content
Since these standard Info cubes are filled from the transactions that you are doing in SAP ECC, you might as well make use of the
common already defined KPI’s. At least doing this will give you some indication that you are collecting and processing the information correctly.
For example if MTBF (Mean Time Between Failures) is not being calculated, you might not be indicating that a breakdown is occurring on
the PM Notification, or in some cases you might not be using notifications at all. This could be an indication that you might need to address / change your processes.
Having the processes in place before you try for global KPI’s is essential. Having part of your organization’s information not being included in the global KPI does not instill confidence in the KPI.
Assuming you are fully using this BW content, and still desire more KPI’s, what next? I will answer that question in another posting.