In order to understand why AI is critical to marketing, it helps to clearly define what Marketing AI actually is. Unfortunately, there happen to be many definitions floating around the web and there is still some confusion about what the term means. An ideal definition should cover all areas of the marketing lifecycle and not just subcategories of the practice. Moreover, a good definition needs to get at the heart of what artificial intelligence does for marketing, since the former is really a tool for the latter.
In his book, Artificial Intelligence Marketing and Predicting Consumer Choice, Dr. Steven Struhl defines artificial intelligence as
anything a machine does to respond to its environment to maximize its chances of success. In the context of marketing, he explains how marketers set their goal as detecting complex relationships we cannot, to aid in our making better decisions.
The above definition of Marketing AI seems like one of the most comprehensive, and does a nice job of summarizing the central idea behind this new, hybrid field. For marketers, the latter part of the definition, dealing with aiding the decision-making process by detecting otherwise undetectable patterns in your data, is where Marketing AI can bring the most value to a company. Marketing in general, and digital marketing in particular, generates a genuine Big Data set to be managed and transformed. The four main characteristics of Big Data are:
The role of data science and machine learning is to tackle all of this data to search for veracity in order to extract its business value. This value can take the form of marketing activation through better personalization of emails, chatbots or better programmatic ad targeting. It can also take the form of deeper consumer insights to improve said marketing activation. Ultimately, Marketing AI seeks to enhance the entire marketing lifecycle, from thinking to doing.
As Peter Drucker famously put it, “The important and difficult job is never to find the right answer; it is to find the right question.” However, your ability to find and ask a good business or marketing question primarily depends on knowing what you don’t know. The more you know about what you don’t know, the more you can discover about what really drives your marketing performance. This is easier said than done when the only tool at your disposal is an Excel sheet or at best, a set of SQL queries to mine your various databases.
Those tools were sufficient in a pre-GAFAM (Google, Apple, Facebook, Amazon and Microsoft) world, when most of the marketing data was sitting in a few Oracle databases and Excel reports. Today, marketers are assaulted with data from a long list of Analytics and Ad/MarTech platforms, each with their own set of complexities and integration challenges.
For example, while most marketing analytics teams won’t admit it, much of the data collected from Google and Adobe Analytics goes to waste. The reason is simple: of the ocean of dimensions and metrics available for analysis, a human being can only process a few at a time. Therefore, when faced with an avalanche of potential columns to analyze, the human brain is unable to transform it into something heuristically digestible, namely knowledge. As a result, your analytics team will either stick to only the few dimensions and metrics they know and recognize (called descriptive analytics or “knowing what we know”) or just vomit out everything they have onto an ugly X-large, all-dressed pizza-looking Excel table report, with shiny rainbow colours and all… The latter is known as data puke. The former leads to skewed or perhaps faulty data interpretations, and shallow insights at best.
Now, you shouldn’t blame your BI or analytics team if this happens. The truth is, digital marketing is a de facto Big Data game, which means humans were never meant to be able to process it all. Even though you paid major dollars to implement platforms capable of collecting ALL the data, expecting your team to process this much information with standard Excel tables is like asking a mouse to eat a whole elephant. This is why GAFAM companies are no longer the only ones investing so much in machine learning. Everyone has Big Data now, to a certain extent.
This doesn’t mean, however, that we no longer need humans. On the contrary. Once the heavy lifting is done by an algorithm, humans are great at taking the results and deriving meaningful understanding and insights. If you equip your smart BI people with the right tools through data science and AI, your company will have the best of both worlds and create a fruitful analytics program and culture within your organization.
For some of you, what I’m describing might still sound a bit utopic. Besides, you might still feel content with the current state of affairs (i.e. the Excel or SQL-only world). After all, if it ain’t broke, why fix it? You may not be convinced of the need to integrate Marketing AI beyond the black box models you’re already using from various MarTech solutions. In fact, doing AI is too complicated and costly anyway, right? As we intend to show in upcoming articles, you’ll be surprised to find out how doing Marketing AI is actually easier than you think. Moreover, stop a moment to reflect on the following questions and imagine a scenario where your biggest competitors have already started to integrate AI into their marketing analytics practice and culture.
Before you answer these questions, take into consideration the following statistics showing the consequences of wasted marketing and business investments, all based on poor or incomplete data exploitation:
For now, the above statistics affect most companies and their marketing departments. Which means you still have time to bury your head in the sand and remain complacent in the face of what I call the Big Data marketing problem. But the clock is ticking. Furthermore, it’s exciting to think that this problem also represents a great opportunity for companies to gain an early mover’s advantage. This can happen by becoming an early adopter of Marketing AI in your business and marketing analytics practice, which is known as augmented analytics.