Press the play button below to listen here or visit the show page at DataDriven.tv

After attending the Global Azure Bootcamp with my son, I decided to drop in and see what bargains there were at the nearby Toys R Us.

I have fond memories of going to Toys R Us when I was a kid. However, I’ve noticed a distinct decline in the store over the last few years. Given the state of their business, I know I am not the only one to see this once great retailer fall from grace. It was a weird feeling walking through a store I had fond memories of and picking through its carcass like a vulture feasting on roadkill.

Naturally, the pundits will point the blame of Toys R Us at Amazon, but is that fair? Is it accurate?

Well, kind of.

Jeff Bezos recently talked about his “one character” emails that forwards to his senior managers when he fields a complaint from the public. These emails not only strike fear into the hearts of the most senior executives at the online retailer, they also spur these leaders to act immediately to resolve the customer issues. What killed Toys R Us wasn’t Amazon: it was Amazon’s relentless commitment to customer service against large retailers who’ve grown lazy and content.

Now that consumers have a choice, they will choose to do business with those companies that provide the best customer experience.

Ok, But How Does Data Fit into this?

It does indirectly. From the article: [emphasis added]

So those customer complaints gives him front-line insights. If all of his data say one thing and a few customers say something else, he believes the customers. 

“The thing I have noticed is when the anecdotes and the data disagree, the anecdotes are usually right. There’s something wrong with the way you are measuring it,” he explained.

Naturally, as a data professional, those two phrases caught my attention. In fact, I see this as a failing in more than one enterprise: the assumption that the BI dashboard knows all.

It might, but it very well may not. If the data tells a different story from what’s going on in the field, then is the field wrong or is the data incomplete?

Put another way: If the map says that there’s a road and there is no road, is the map wrong or is reality wrong?

Clearly, reality is the ultimate arbiter and if you’re business is retail, then your reality is customer perception, not what the pretty graphics on your KPIs say.

In this video from the Ignite conference last September, watch the latest additions to the Cognitive Toolkit, which offer a Python API, as well as a GUI to have a non‐disruptive experience from data load through operationalization with all the steps in between.

The goal is to support classification, object detection and image similarity use case. This is a work in progress. In this session, they demonstrate only a classification pipeline.