3 Ways to Use Machine Learning in Your Marketing

Blog Post
August 29, 2017

Machine learning is essentially giving computers the power to learn from experience, very much like humans do (learn more about what machine learning is and who is already using it well). Marketers can take advantage of this technology in innumerable ways, but there are a few capabilities that will really help to supercharge marketing performance:

  • Personalization
  • Attribution
  • Sentiment Analysis

Machine Learning for Personalization

Most consumers today expect companies that market to them to “know them," to know what needs they have, what their goals are, and the problems they are trying to solve. In short, consumers want to feel they are special. This is where machine learning does the heavy lifting, capturing patterns quickly that are in turn used to create more personalized offers and experiences.

The more you understand about your customers, the better you can serve them, and the more you will sell. This knowledge and understanding is the groundwork behind marketing personalization. A growing proportion of consumers expect the benefits afforded by machine learning, even though they do not really understand or even know that it is running in the background.

When a consumer is served an ad, he or she expects that ad to be relevant, meet a need, or solve a problem. When consumers make an online query, they expect quick, insightful answers. People want and expect personalized experiences at every touch point along the customer journey. Machine learning makes this happen by enabling decisions in real time regarding which experience to serve each consumer, leveraging data points generated by not only those consumers, but all other consumers as well.

Let’s now link this concept of “personalization” to traditional segmentation solutions that have been around for decades. In my experience, when marketers are tasked with a segmentation project, they will typically create three to five segments based on some combination of data points: transactions, value, demographics, psychographics, needs, attitudes, etc. These segmentations and corresponding profiles that are created are in turn used to better inform and optimize the marketing strategy. However, traditional segmentation solutions are not scalable. Let’s assume that a company has just implemented a segmentation solution that contains five segments and corresponding segment profiles/personas. Can the marketer really create truly personalized experiences and content based on insights from the segmentation project that generated five unique profiles/personas? It’s not possible!

Retailers in particular are really pushing hard in the personalization space. Huge retailers like Walmart carry millions of different products with incredibly large inventories and are able to adjust that inventory in almost real time depending on what’s happening in the market. For example, if a huge snow storm hit the southern part of the U.S., Walmart can very quickly shift the inventory in certain stores and online to items they know customers will buy, (e.g. sleds, snow shovels, snow hats, etc.). This results in a more personalized and relevant experience for consumers by leveraging the power of machine learning and artificial intelligence.

Machine Learning for Attribution

Advertising in today’s turbulent marketing media landscape, where individuals are constantly bombarded by numerous digital ads and offers, requires real-time tools that can quickly identify patterns and make deductions, which in turn can be applied to future deployment of media.

From an attribution perspective, machine learning can be a huge boon to data processing. Past experience, knowledge, or simple affinity toward a certain type of marketing media can quickly bias how a human attributes desired outcomes to various marketing stimuli. This is particularly true in last touch or any type of rules-based attribution. Last touch attribution is when all credit for the desired outcome (an online purchase) is given to the last known touch or interaction. Whereas rules-based attribution will spread the credit across a handful of known touches and interactions within a defined tracking window (e.g., 25 percent credit given to the first touch and the remaining 75 percent given to the last two touches in a 60-day tracking window). Machine learning takes the human bias out of the equation.

Leveraging machine learning is critical for enabling the consumption and interpretation of large amounts of data and doing so in a completely unbiased fashion. Machine learning allows massive amounts of digital data to “speak” in order to weigh the results of each touch, interaction or micro moment within the customer journey.

Machine Learning for Sentiment Analysis

Machine learning is also a terrific tool for sentiment analysis, which is the process of using text and linguistic analysis to determine the emotion behind a series of words. Think of it as mining data to determine a consumer's opinion about a given product or service.

For example, let’s say an online game creator released detailed information and features of a new game along with a limited play version to initially gather potential buyer’s opinions. Gamers would no doubt take to social media to express their opinions…good and bad. The game creator could then use the power of machine learning to understand the flood of opinions and leverage the understanding to make any final changes to the game prior to the release date.


Machine learning capitalizes on advances in computing software and the sheer processing power of computers. The possibilities for marketers are endless, but a smart place to start is by using this technology to improve customer experiences through personalization, improve attribution to optimize your marketing spend, and analyze sentiment in the market.

You’ll want to stay tuned for my next post for some practical tips on getting started with machine learning.