Ending D4DS

@timsuther recently tweeted the most poignant article I’ve read in ages on the correct way to use data science. Dr. David Stephenson’s “150 Data Scientists and still no business value?” blog post presents the most important data concept to understand as 2017 begins: business value needs to drive real analytics vs. Data for Data’s Sake (D4DS).

From one of my favorite websites: Spurious Correlations

D4DS is the antithesis of business value and a perversion of Deming’s “if you can’t measure it, you can’t manage it” philosophy. Many things can be measured that have little to no value. My personal pet peeve is social media Sentiment Analysis. Guess what? Lots of people say a lot of silly things online that generate petabytes of data. For most companies, time would be better spent enabling social channels for targeted, outbound communications to social custom audiences vs. the time and expense of a data science exercise for dubious correlations. For some companies, social Sentiment Analysis is a game changer, such as pure brand research. However, in every case the business need is well thought out in advance of a scalable production solution.

What can you learn from this? Nothing. There is no reason this “sentiment” should be measured and added to a database.

Measuring anything takes time. Time is the true gold standard of a business. I would only add the following to Dr. Stephenson’s exceptional suggestions for managing a data science program: start with a clean foundational data layer and build from there. Take the time to understand the data you have and organize it in such a way that many in your organization can explore, build, and create insights.

Please note that this is not a prescription for any particular brand or method of data management: your business is your business with unique needs. The real value will come from the effort of organizing and explaining your data. Then you can measure things of value. Start small, iterate, and know the hard work will pay big dividends when the data does get big!

Or not — you can always jump in a Data Lake!

As a partner at SSB, I do (shamelessly) recommend our SSB Central Intelligence data management platform for anyone who needs to build a data layer to run their business. It measures things of value. Value is what matters.