By Stephen Sklarew, CEO

#1: Early adopters are winning

Amazon, Google and Facebook are relentless innovators and show no signs of slowing down. Their ability to stay relevant and drive huge gains in recent years is built on large and early investments in collecting massive amounts of data and learning how to successfully apply machine learning to monetize it.

But what about applications in more traditional industries?

Companies across industry are touting significant returns on investment:

  • Airbus uses machine learning to help detect clouds in satellite images, decreasing error rates from 11% to 3%.
  • A food manufacturing company in Japan, Kewpie, uses machine learning to detect defective potato cubes. Its accuracy, at par with human inspectors, saves $100,000 per production line.
  • Case Western Reserve applies deep learning to automatically assess cancer risk at a fraction (1/20) of the cost of current genomic tests.
  • Deschutes Brewery uses machine learning to determine when to stop one part of the brewing process and begin another; saving significant time producing beer and reducing waste.

The number of success stories is accelerating.  Imagine how many companies are starting to implement big data and machine learning that don’t want to let their competitors know.  If you aren’t convinced, check out MIT Technology Review’s recent survey results.

#2: It takes time

Machine learning is an entirely new way of thinking.  It requires a radical mindshift from what most teams do today.  In the traditional approach a company sets business objectives, analysts collect requirements, programmers code rules into applications, and testers verify that the applications meet the requirements.

Successful machine learning implementations also rely on business objectives; but, before building a system, there needs to be available data, a deep understanding of data, and experienced data scientists. The long pole in the tent is often making the data available — collecting, cleaning and preparing it for analysis. Choosing the right algorithm and training up an accurate machine learning model properly isn’t easy either especially if you don’t have experienced data scientists.

Not only does your organization need to learn how to approach designing systems differently, but you need to get your data in a good place, hire experienced data scientists, and build accurate learning models. The longer you wait to get started, the further behind you are from your competition.

#3: You could be saving time and resources

To stay competitive, your organization needs to optimize its productivity.  The good news is, with a little investigation, you’ll find lots of repetitive manual tasks that can now be automated with minimal investment. Machine learning can automate many types of manual tasks so your team is more focused on the work that keeps your company relevant instead of drowning in tedious, brain-draining manual work.

For example:

  • Almost everyone in the business world struggles with their calendars to schedule meetings — many services like are tackling this problem with machine learning..
  • Radiologists now have machine learning systems to help them pinpoint images they should focus on, instead of doing all analysis with the naked eye.
  • Media companies are still using librarians to manually create taxonomies, map taxonomies, and link content. We’ve used machine learning to automate this work for our clients.

The Bottom line

If you think back to the mid-1990s or early 2000s, few would have guessed that Amazon, Google and Facebook would be so successful. As they continue to extend their dominance, entire industries are changing.  This change is driven by these pioneers’ incredible ability to collect and leverage data. Implementing machine learning has opened their doors to a brighter future.

Now that big data and machine learning are becoming more and more accessible to everyone, it’s time to future-proof your company.  To do so, you need to: 1) Optimize your employee productivity; 2) Own, harvest and evolve a unique set of data assets; 3) Apply machine learning. The longer you wait to make this shift, the harder it will be to catch up.