Marketing activities create a deluge of data. Today’s marketers face, at times, an almost overwhelming number of challenges when it comes to deploying truly personalized, choreographed and impactful communication streams across the customer journey. For every $1,000 spent, marketing activity can generate literally millions of data points in the form of impressions, clicks, responses, conversions, revenue, and sentiment data points. So how does one make sense of this plethora of data? This is where Artificial Intelligence and “machine learning” swoop in to save the day.
In this post I’ll be explaining what machine learning is and what it does. My subsequent posts in the series will cover:
What Is Machine Learning?
Very simply, Machine Learning is a form of Artificial Intelligence (AI) that is used to solve problems by finding patterns, deriving insights, and making predictions from tremendous amounts of data. It enables computers to churn through this data and to learn and improve without explicitly being told to do so.
Let’s say an analyst wishes to build a predictive model to score and rank individuals on their likelihood to buy product x. Roughly speaking, the analyst will follow these steps:
- Collect and compile the data
- Perform an initial exploratory data analysis to identify potential predictors
- Build the model
- Implement the model in a centralized customer database
- Structure a test to validate the model in the “real world” via a marketing campaign
- Collect the results and evaluate the performance of the model
- Utilize the results and corresponding new data points to rebuild the model to hopefully improve upon the results generated by the initial model
- Repeat steps 4) through 7)
As you can see, it is a manually intensive process that mainly consists of data manipulation and manpower from analysts and data scientists to find the insights and improve the predictions going forward. Machine learning, on the other hand, leverages the massive processing power of computers. Their objectivity allows them to see patterns in tremendous amounts of data that the biased human mind cannot, and then apply those insights to determine how new data can be used to accurately predict results.
Algorithms and Anomalies
Machine learning is terrific at discovering anomalies in data and finding the needle in the haystack very quickly. Huge tech-driven companies such as Netflix and Amazon utilize these highly intelligent algorithms. They predict products you may want to purchase based on comparing your previous purchase history, online search patterns, and a plethora of other data points, to the data points of millions of other users to determine what you may wish to purchase or what movie you may want to watch next. Yelp hosts many millions of photos using machine learning to identify and understand what those photos contain, and then to classify them. Google uses AI and machine learning to identify and eliminate spam from your primary email inbox. In fact, in the digital age of today, consumers have come to expect these benefits of machine learning.
Machine learning also does a tremendous job of grouping like things together or creating “bins” of data. I’ve found over the years that the human mind can wrap itself around about three to five data elements at most to find meaningful patterns in the data, create basic predictions, or to create basic segments. After that, the average human brain starts to let us down a bit… plus, it is almost impossible for a human to look at data from a completely objective point of view without human bias.
The amount of data marketers are now able to collect can be overwhelming. Computers can't yet replace humans, but they are far superior at sifting through and assessing raw data. Companies from Pinterest to Tesla are harnessing the power of AI to derive insights from millions of individual data points—insights that the humans on your team can then use to implement truly meaningful communications streams.
Hear more about how companies like this are using machine learning to their advantage in my next post: Machine Learning: Separating the Hype from Reality.