WHY IT MATTERS: Digital Transformation

This is just the beginning of insights we will get as we analyze the vast amount of data now being produced by digital devices, large or small, fixed or mobile.

Interesting raw data. As we get closer to industrial internet, MDM, quantified self, etc. there is going to become an even more important data flood...

The post dates back a little bit but GE continues to focus on this new field they call industrial internet.

Useful data tidbit: a single airplane jet engine captures 5000 data points every second and generates GB of data on every flight. The impact of proper analysis of this data can result in tremendous savons and improvements. For example, a 1% improvement in efficiency in a gas turbine, over 15 years, results 66B in savings! 

GE's challenge is not to capture this data (they do already) but it is to analyze it. IDC estimates that only 0.5% of the world's data is being analyzed. New tools and techniques related to BigData (hadoop, noSQL, etc.) can provide some answers.

Additional reading:

Just 0.5% of the Worlds Massive Trove of Online Data is Being Analyzed via @IDC http://sco.lt/8LRfhR

Retailers have been doing forecasting for a long time, predicting from previous sales and market conditions (and their best judgement), what to order for each one of their stores. When you have hundreds of stores, this is not a simple tasks, and it often requires years of experience - if you want to do it well. Because forecasting errors are very costly, resulting in lost sales or products that stay on shelves or racks and must be transferred to others stores, or worse, discounted at the end of the season.

I've worked on systems - from simple excel spreadsheets to more complex applications - that automate the predictive aspect and provide tools to facilitate the forecasting process. But never were the tools using invidual client profiles  to make the prediction. They always used global store-level or regional sales numbers. 

So, at least in theory, established retailers can have a leg up.

But Amazon may (again) shock retailers into a digital transformation, bringing their individual client sales data into account when doing the forecast. This will require retailers to include individual customer transactions and possibly their "wish lists" and online search information, to include in your forecasts. Data is there but few reatilers have the systems or the integration to bring this data in a timely manner to the "forecasting team" desk.

Amazon has been doing this with "replesnishment" options where they send you products on a regular schedule, so you don't have to "shop" for them: they come at your door at a predefined interval you control. And they give you rebates for sticking to your schedude (predictability is invaluable in a supply chain!).

If you do, then your forecast will undeniably be better. If you don't companies like amazon, with lower overhead, will send your clients products before they even consider going to store to see what's on the shelves!

I have yet to put those reference designs / architectures to the test but the enterprise architect in me finds it a good thing when I find patterns like that!