Being Bold and Big with your Big Data Pilots


Author: Joseph Baird

I get it. We live in a corporate world where failure is a stigma.  What if we screw up?  What if I lose this position?  What will people think? We also live in a corporate world that is starving for innovation.  The Revolution is not being televised on channels 2, 3 and 7!

In years past, large corporations had research labs that consistently produced innovation.  In the past decade, Google had its famed 20% time.  It provided an environment to experiment and by its very nature fail and learn and fail and learn and succeed.

In this domain of Big Data, Big firms have a wealth of information and are uniquely positioned to create innovation.  But there is also an immense amount of noise in the modern, large company.  And cutting through this noise requires making a loud splash with the results of new work.

In my work with a wide range of large companies, I have seen the starts of many pilots.  Some started as technical proofs of concept.  Some started as a little test by the marketing group’s middle managers.   These too often than not they were curiosities.

They may have helped a technical resource learn a new marketable skill or a business person to create an interesting PowerPoint, but more times than not the efforts languished.

Now, though, is the time for courage by executives to create a safe environment of bold thinking.   As with all analytically related problems, start with the questions you want to answer.  And make it questions that really matter to the business.  What would be real game changers if we could know with more certainty?

And then set the expectations with your fellow executives that you are sponsoring this bold experiment and you would like their support. The beauty of today’s technical world of open source and SaaS players is that big, meaningful pilots can be done for small prices.

Get your skunks together and make it happen. And then publicize the results whatever they may be. Remember, the learnings are the most important thing.  Chances of succeeding perfectly are very rare and in fact probably undesirable. Your job is to drive learnings not create a production ready system.

So in closing, a few tips for the executive willing to sponsor the Big Question:

  • Rent it on a private cloud and don’t get bogged down in IT roadblocks or expensive vendor software purchases.
  • Create a big hypothesis that is dogging the business.
  • Take advantage of the new class of data blending software.  Historically, data prep has taken the bulk of project time.  Leapfrog solutions that learn to stitch data quickly are available.
  • Visualize it and share it widely.  We have found that sharing the interim results with a wider audience can produce advances in data quality as well as insight.  Caveat the heck of it and set expectations with your audience. THIS IS AN EXPERIMENT!
  • Isolate your lab and then take the keys away from legal and procurement.  Once you have established general PII protection, you need to let unbounded design thinking occur.

The choice is really up to the leaders of large companies.  Invest now, or pay very large sums later.  $4B for Snap Chat?  So go BIG with your big data experiments.

Reblogged from Original Post:

Be Bold, Be Big with your Big Data Pilots

OrigiN

Advertisements