The Case for Machine Learning in Digital Marketing

Not far from now, you will have hard time in figuring out whether this  blog is written by an machine or a human! Machine learning, the recent buzz word in the tech industry have started closing the intelligence gap between humans and machine. And its making inroads in many sectors of industries , which are solely dominated by humans.

Machine Learning in DIgital Marketing


Deep learning, the newcomer in machine learning  have  recently even outperformed humans in terms of  recognizing complex objects in images. What differentiates machine learning is that  it is not the usual programming that we are taught in school, where we instruct the machine how to respond by a specific set of rules(algorithm)  for every instance of input it gets from the environment, but rather it tries to replicate the human intelligence which  extracts the useful patterns or signals from the huge amount of data scattered with noise.  This is similar to the method proposed by Jeff Hawkins to create the true artificial intelligence in his classic book ‘On Intelligence’.

How Machine Learning works

In most of the machine learning techniques, the method of learning is similar to how humans learn in the real world. There is a stage called training where the machine makes all sorts of mistakes(error) while trying to predict and it constantly tries to reduce error between the actual and predicted values. The beauty of it all is mathematically it boils down to an simple optimization problem of minimizing error. If we see humans , from childhood we first learn a skill by making countless mistakes and we build expertise through practice. The same applies to machine learning, train them with right amount of relevant data, over time it will start challenging humans in terms of accuracy.

How machine learning works
Source : Machine Learning in Simple Words

Paradigm shift in marketing

Before the era of digital marketing, the marketers had few data points and completely relied on one of the  powerful features of human brain to find leads. i.e intuition  evolved from millions of years of evolution.  As we moved to digital era, things have changed. The amount of data points that a typical marketer has to interpret every day is in the order of millions which is  way beyond the limitations of an average human brain. This is where strength of machine learning  lies, the ability to extract patterns from huge volumes of data which have high velocity and variety.(i.e Big data).

Exponential growth of data










Let’s take one of the important activities of marketing, the prioritization of the leads. The way a typical marketer does is by scoring the leads by assigning a set of points for each characteristics of lead. For example if  the lead  job profile is CTO assign +10 points, if lead is intern assign -5 points, if lead downloads a whitepaper assign +3 points. As you can clearly see there is lot of subjective bias and guesswork. Added to that these rules has to be regularly updated or refined manually over the period of time. But instead if we use the machine learning algorithm to extract meaningful relationship and patterns from the historical sales data, there is no need to worry about the update, as most of things are automated and whenever a new deal is closed the model is refreshed automatically with new set of data points. It also identifies non-linear relationships between data points which are normally lost by these human built rules.

We know  that out of large volumes of  marketing data that we gather in this 21st century, finding the fraction of data which is  useful  and relevant  is equivalent to finding a needle in the haystack. There is no doubt that one of the successful ways  of  finding our golden needle at proper place and time lies in how we properly utilize the machine learning techniques in the domain of digital marketing.

This blog is written by Sriram Ramanathan from BRIDGEi2i