Over the past 22 years we’ve seen many organisations embark on internal IT projects. Of all those projects only a scant few could have been classified as being truly successful though. Of the rest:
It’s actually only a small proportion that have delivered good business benefits in the end. Just a few that have not only paid for themselves but also generated a return. What is it about these projects that has made them successful? Is there a common theme? Yep, you’ve guessed it – each of them has been business led.
In the last few years more and more of the failing projects we’ve seen have been BI projects. There’s good reason for this – everyone has the data bug. All the large consultancies are telling us all that we’re creating more data than ever before, the value of ‘Big Data’ is apparently through the roof, and we’re working with companies that are striving to add ‘data’ as an asset to their balance sheets. We’re entering an age where information is perceived to be more valuable than our people, our customers and even our products and services. Everyone is rushing to build gargantuan data warehouses or buying data blending tools and we’re all looking to hoover up as much information as we possibly can.
There’s a fundamental problem with this though. As Nate Silver tells us in The Signal and the Noise, more data doesn’t mean better decisions. In fact, more data is, in the vast majority of cases, just confusing the issue. The more data you have the faster you approach chaos in every sense of the term.
This is surprisingly true even as we enter this new world of machine learning. Many machine learning techniques are built around looking for the right data, seeking out that tiny signal in a sea of noise. Most of us are standing on the outside of all this new technology looking in right now, so more on that in a future article.
So without piling all of this data into some deep learning, neural network and hoping for the best, how do we sort the right data from the wrong data?
Well, instead of starting with all the data you have and attempting to piece it together in order to identify some correlations and then looking for causality, we flip the whole process on its head. Start with the questions that are important to your business. Instead of leading with technology, lead with business requirement. It sounds so simple but you’d be amazed how often IT leads the charge.
Chief Analytics Officer Spring 2017
15% off with code MP15
Big Data and Analytics for Healthcare Philadelphia
$200 off with code DATA200
10% off with code 7WDATASMX
Data Science Congress 2017
20% off with code 7wdata_DSC2017
20% off with code AIP17-7WDATA-20