The Future of Digital Marketing – Machine Learning

July 20, 2021 Peter Iuvara

An abbreviated history of the Digital Marketing Landscape:

  • 1991 AOL
  • 1994 First Online Display Ad
  • 1995 Yahoo! Search Engine
  • 1996 Email Marketing
  • 1997 SEO
  • 1998 Google
  • 2004 Responsive Web Design, Facebook
  • 2005 Google Analytics
  • 2006 Twitter, Marketing Automation
  • 2007 iPhone
  • 2008 Facebook Ads
  • 2010 iPad, Instagram
  • 2011 Pinterest, Snapchat
  • 2014 Mobile Surpasses Desktop Users
  • 2015 Wearables
  • 2016 Internet of Things (IoT), Online Spend Surpasses TV Spend
  • NOW …

A Digital Marketer has access to an incredible amount of data and channels, more than ever before. Social platforms alone represent over 1+ Billion unique users globally, and that’s just Facebook; and, with Data, well, here are some facts and predictions:

  • More data has been created in the past two years than in the entire previous history of the human race
  • Over 1+ billion people use Facebook every day
  • By 2022, we will have over 6.6 billion estimated smartphone users globally
  • By 2022, there will be an estimated 11.5 billion smart connected devices in the world, all developed to collect, analyze and share data
  • By 2022, an estimated $400 billion will be spent on public cloud usage
  • At the moment unto 73% of company data is not be used

From a financial perspective, the value is tremendous:

  • Estimates suggest that by better integrating big data, healthcare could save as much as $300 billion a year
  • For a typical Fortune 1000 company, just a 10% increase in data accessibility will result in more than $75 million additional net income.
  • Retailers who leverage the full power of big data could increase their operating margins by as much as 60%.

Current Practice

The reality is the use of data is far less optimized than the collection. We have built systems and applications that are excellent at collection. Many companies have undertaken projects to build out their Big Data and Business Intelligence functions, and have gained some operating efficiencies along the way. But practically every company has only scratched the surface of the potential here. Let’s take a look in practice.

Tools: website, content copy, graphics and a call to action

Ask: we need to sell more yoga mats

Sounds simple enough, right? So what do we do? Well most companies I speak with run an A/B Test. Test A has copy, images and a theme (colors and general aesthetics) with the “Buy Now” call to action. Test B has the same call to action, but different copy, images and theme. The test is run for 1 week, and programmed to serve up Test A 50% of the time, and Test B 50% of the time. The data is collected and reviewed in terms of which Test produced a more favorable outcome – more call to action conversions. In other words, we backing into designing a Decision Tree to find out which branch(es) yield the highest probability of success.

The test is very simple in nature, the goal is to obtain the highest conversion; let’s call that variable x. The test condition is variable y, which was the “look and feel and copyrighting.” But let’s look at this a example now a little deeper and introduce Machine Learning.

Discriminative Model

Discriminative Models or conditional models, are a class of models used in Machine Learning to model the dependence of a variable y on a variable x. As these models try to calculate conditional probabilities, they are often used in Supervised Learning.

An example of this would be Logistic Regression used to predict whether a person will purchase a flux capacitor (variable x) based on what is visually displayed to that person (variable y). The key here is prediction – very important.

Future Practice

Machine Learning in conjunction with an Enterprise Content Management System and Marketing Automation Platform can create an automated process in which the test in automatically conducted at an individualized-level with a degree of certainty far greater that chance can ever be. Here is how – data combined with Machine Learning.

  1. Start with the data, think about what is possible, and what is collected on the consumer side.
  2. Introduce the Targeting Algorithm and Measure. Perform the test, but now it’s not just an A/B test, it’s a billion test-case test for theoretically every consumer there is data collected for. Data around preferences, interests, gender, location, previous purchases, age and other data points can all be used to make the call to action ad very targeted.
  3. Optimize through Machine Learning. Over time, conversion rate performance can be stored and used as benchmarks both at an individual level and collective level.

This application of Machine Learning is really geared towards optimizing problem-solution processes. The real power here is in leveraging a machine to perform billions of operations in a logical manner to arrive at a single, ideally, best decision.

Many companies are already using Machine Learning, take Google for example. When a misspelled search is submitted, it is using Machine Learning to detect the misspelling and attempt to automatically correct it, usually successfully.

Marketers and marketing automation platforms are starting to leverage Machine Learning as well. Some brands are using email marketing more efficient by leveraging Machine Learning to analyze consumer behavior and sentiment to determine when email delivery is most likely to draw engagement and conversions.

Behavioral, sentiment and contextual data is becoming more available to brands, not because it is newly created, but because applications are starting to now leverage Machine Learning to make Digital Marketing more effective. Without Machine Learning, all of this data would be useless because marketers would be unable to put that data into context and drive insights from the information without Machine Learning. The sheer volume of data would be to arduous a task to gain insights from otherwise.

Evolution of Machine Learning in Digital Marketing

In addition to the obvious elements we are able to target based on now, here are some additional uses, predictions and future trends we see coming:

  • Weather data and analysis for targeting
  • Competitive analysis for product/service benefits
  • Storytelling automation – imagine that, where Machine Learning can not only target and dictate optimized copy elements, but personalize a story based on combining brand products/services with personal information. Imagine a retailer of shoes, that has access to your calendar. Not only would the shoes be displayed, in your favorite color by default, but it can also orchestrate a story including upcoming events you are planning to attend in the near future and how these new shoes would be ideal.

Contact Us For Help With Digital Marking Automation

Our team has been working with Digital Marketing Automation systems since 2010 and have seasoned experts that can help you begin with this cutting-edge technology. We would love to hear from you; contact us to setup a working session to start your journey.

, , , , , , , , , , , , , , , , , , ,

Leave a Reply