Machine learning: 3 places news media companies should start

By Kirk MacDonald

Pyrrhonian Partners

Toronto, Ontario, Canada


If there is one thing we are not very good at, it’s embracing change.

Our inability is magnified when we work in a legacy industry where unwritten “how-to” rules have been handed down to the next generation of leaders. Those leaders are generally unequipped to cope with today’s intense disruption where rules are being rewritten on the fly by technology.

Machine-learning technologies can assist media companies with data analysis, decision making, and more.
Machine-learning technologies can assist media companies with data analysis, decision making, and more.

By definition, disruptive innovation is transformation. This means the transformed doesn’t look anything like its previous self. Disruptive innovation is not a transition with incremental improvements unlikely to be readily recognisable.

Or, put another way by Silicon Valley guru and Zero to One author Peter Thiel:

“Unless companies invest in the difficult task of creating new things, American companies will fail in the future no matter how big their profits are today. … Today’s best practices lead to dead ends; the best paths are new and untried.”

Through experience and learning, I’ve developed a thesis on why there are so few examples of successful disruptive innovation:

  1. There is little or no value placed on organisational learning, which results in a lack of awareness.
  2. Without a formal learning framework, we lack understanding.
  3. Without understanding, we lack the capability to solve problems, apply new approaches, innovate, and become disruptive.

The result: Decline. Slow and painful.

Technology has the potential to change that and be a positive disruptive force for media companies, or any other legacy industry, with a commitment to learning and change.


Here are three examples of related machine-learning attributes, assuming companies have the infrastructure and commitment in place:

  • Data and analytics.
  • Decision-making.
  • Predictive analysis.

Input is the data. Output is the analytics that determine decisions and foresight. This is the new model for decisions made in competitive organisations.

Data and analytics

Use of data has been discussed for years, but applied data and analytics remain elusive. In any computer network, software captures every keystroke and image and stores the information in a master console, whether it be on a server or in a cloud.

Over time, it stores massive amounts of bytes that provide employee, customer, and competitor data that can be used for significant performance improvements by analysing patterns and trends. This becomes a data science exercise.

(A good understanding of data and analytics can be gained by reading Michael Lewis book “Moneyball” or watching the movie starring Brad Pitt and Jonah Hill.)


Improvement in this area has a direct impact on the bottom line. The patterns and trends are free of emotional bias and represent black-and-white options on which to base a decision.

There is a reason computers can beat human beings at chess. Better decision-making that comes in the form of Artificial Intelligence generated from machine learning overcomes human bias we face at the crossroads during decision time.

(A good understanding of automated decision making can be gained by reading this December 2016 Guardian story, featuring Bridgewater and Associates, the largest hedge fund in world.)

Predictive analysis

This is already happening at many companies that have built software enablement tools to feed machine (computer) learning and AI. While SEO and referral engines are simple examples of predictive analysis, the company to learn from is Salesforce.

Salesforce is far and away the leading enterprise solution for customer relations management (CRM) and is using the Internet of Things (IoT) to provide customers with probability insights you can’t get anywhere else, along with the organisational learning that comes with the platform.

(A better understanding of predictive analysis combined with machine learning can be found by reading INMA’s blog post on Schibsted’s experience increasing telemarketing conversion rates by 540%).

These areas would be good places to start the disruptive innovation journey.

About Kirk MacDonald

By continuing to browse or by clicking “ACCEPT,” you agree to the storing of cookies on your device to enhance your site experience. To learn more about how we use cookies, please see our privacy policy.