There seems to be ongoing confusion in the market regarding the use of the term “analytics” . Wayne Eckerson made some great points in a blog last year titled “Bastardizing Analytics”. A short aside - as an Australian I am naturally drawn to any appropriate use of the term “bastard” as we tend to use it as often as possible in as many contexts as possible, so good work Wayne.
The confusion hasn’t been helped by vendors using the term analytics to cover anything from simple data filtering to extremely sophiscated modeling technology. I think an easy way to view this is that “analysis” is a process whereby a person uses tools, experience, knowledge and their intellect to investigate a situation and determine the cause and impact of that situation. There are a number of tools that can be used in this process, including:
- data reports and visualizations;
- filtering, sorting, ranking, drilling and breaking data down to obtain more detail; and
- applying analytics to extract more information and understanding from the data and to determine corrective actions and predict likely future outcomes.
Each of these in turn is more sophisticated. So given analysis is a process and analytics is tool, what constitutes analytics? When looking extract more useful, actionable insights from data there are a number of capabilities that can help, including:
- using statistical algorithms to apply trend lines to historical data to gain a better understanding of the overall direction a particular metric is heading;
- projecting trend lines forward (with or without error ranges) to gain an understanding of what might happen if no action is taken and the current trend persists;
- applying variance analysis (absolute, relative, waterfall etc.) to see the change from one reference point or in one metric relative to another;
- using scatter charts to determine the degree of correlation between multiple metrics;
- using Pareto charts and histograms to identify the major contributors to an outcome or issue;
- using control charts (X-bar and R-bar with multiple standard-deviations) to determine how much variability there is in a particular metric to help identify out-of-control and systemic issues;
- using scenario analysis and modeling to determine the overall impact of one or more specified changes (either on a historical or future value) to determine the most effective actions to take to improve overall results.
Obviously these is a degree of sophistication to the above and you may be wondering whether analytics tools can only be used by business analysts or statisticians. I believe the answer has to be absolutely not if Operational BI is to become pervasive and help improve operational performance and business results. Analytic tools should be easy to use, understand, apply and consume as part of the every day decision making undertaken by many people within any organization.
This can be achieved by offering a selected set of analystic tools that can be applied to a metric or a particular metric visualization depending on the characteristics of the metric itself. For example being able to apply a control chart to a revenue metric probably makes no sense while having the option to apply a control chart to product quality or defects would provide valuable insight. It can also be achieved by allowing people to specify an analytic action (for example a scenario analysis) within the context of the data they are viewing and the decision they are trying to make.
For example, as I look at production output I might take the following path:
- Drill into production output and see the contributors to production by facility and then by production line;
- Select the aspect or dimension of most interest and drill into the factors impeding production (maybe in a waterfall variance) such as turnaround time, breakdowns, operator breaks, quality rejects etc;
- Then within that context apply a scenario such as “what if we were able to decrease the reject percentage” on this production line within this facility
- This would be a natural flow to solving the issue of insufficient production output.
In summary analysis is a process that can use multiple tools of which analytics is one, and the value of analytic capabilities is maximized when they can be applied across the organization to help everyone making daily operational decisions to make better decisions, more quickly.
{ 0 comments… add one now }