Artificial Intelligence in the Marketing Realm
Understanding how AI can expand the marketer’s toolkit.
Welcome to the age of buzzwords. To be more precise, welcome to the age of tech buzzwords. Blockchain, internet of things, decentralized finance, non-fungible tokens, artificial intelligence, machine learning, deep learning, quantum computing, software as a service… To make things worse, all of these have acronyms you keep forgetting all the time. I’ve found that the best way to understand and digest these massive concepts is to learn about use cases in specific domains. So the following words will hopefully answer some of these questions. How is artificial intelligence (AI) useful in marketing? How can marketers realistically integrate it within their toolkit? What are some examples where AI techniques shine?
What is AI?
Search around and you will come across different definitions, but I really dig IBM’s approach:
“At its simplest form, artificial intelligence is a field, which combines computer science and robust datasets, to enable problem-solving. It also encompasses sub-fields of machine learning and deep learning, which are frequently mentioned in conjunction with artificial intelligence. These disciplines are comprised of AI algorithms which seek to create expert systems which make predictions or classifications based on input data.”
Marvelous. As you can tell, the first thing we need to do as marketers is to move away from the science fiction robots with General AI capable of exterminating the human race. We are nowhere near that scenario. Where developments are taking place today is in what is known as Functional AI. Bear in mind that AI applications used today are more capable than humans but only in very specific tasks. I cannot stress that enough.
Machine Learning (ML)
ML is a subset of AI. Think of it as the tool that’s most widely used today for practical applications. ML derives patterns from data to make predictions. It establishes rules among data points and their interrelation. We say it is able to “learn” because it can reassess its models and improve as new data comes along. To train and optimize ML models, we need lots of data. And it’s not just about quantity. Data should be treated as the most precious resource. It has to be clean and reliable. We need to understand how our data is built, what it holds, and how we can expand it. As marketers, we must learn how to interpret our data and understand the qualitative aspects of how it’s feeding into our models and helping us derive conclusions.
It’s also convenient to get familiar with three key areas of ML. In simple terms:
- Supervised learning: when you know all data inputs and map them towards a known output. Give a machine a ton of examples, they will learn from the data.
- Unsupervised learning: when you ask the machine to find a correlation in the data that we didn’t previously know about. This is a powerful ways to find patterns and clusters.
- Reinforcement learning: give the machine a task and a goal and it will make changes that improve the outcome based on its own actions.
Putting it together: where ML shines
ML is excellent at dealing with dimensionality and cardinality in the data. Dimensionality refers to all the attributes (dimensions) of an occurrence — think about columns on an excel sheet. Cardinality refers to the number of options within each dimension. So if you’re looking at a database on houses, then location/price/size/rooms/has garage/condition are all examples of dimensions. Then “price” has infinite cardinality whereas “has garage” has only 2: yes or no. So ML is suitable for dealing with large and deep data sets and finding patterns in the data that humans couldn’t possibly do.
Use cases in Marketing
First, a few questions. Do you have lots of available data on existing or potential customers and how they interact with your ecosystem? Does this data have an overwhelming amount of combinations? Is it clean and well labeled? Do you have clear goals on what you would like to use this data for? Are there repetitive and high-volume processes that can be automated? Then you might have an AI use case. Examples are endless, here are just a few thought starters.
Think about your new ad campaign. You would use ML for finding correlation in data, segmenting audience, clustering people, and locating anomalies. It’s suitable for ranking, sorting, finding patterns, and prospecting look-alikes. All of these valuable for improving your campaign effectiveness as you go. Take it further and think about programmatic advertising, where AI automates ad buying via algorithmic targeting tactics that optimize content, placement, and consumer audience. Nowadays it’s possible to choose the desired outcome of your campaign directly (more views, more subscribers, more sales…) and let the algorithm do the rest, instead of inputting what you think is the best target audience.
You could leverage the power of AI for natural language processing (NLP) that can analyze voice and text and identify sentiment as well. Think about customer service chat bots or AI sales reps guiding your consumers through the funnel.
You could gather data on your customers in order to create personalized experiences and recommendations at scale; those that optimize conversion, retention, and lifetime value. ML models can also analyze their attributes and behavior as they interact with your channels and offer them the most relevant content or products, at the best time.
Here are some other uses cases you might want to research more deeply: sales forecasting, AI-powered customer insights, content generation, speech recognition, dynamic pricing, automated email content curation, and image recognition.
Takeaways and thought starters
As a marketer, I encourage you to discover those areas of your business that are ripe for AI implementation and will make a difference in the eyes of the consumer. Look for repetitive tasks or areas where there are lots of data points from which insights can be extracted. Figure out how these insights can be leveraged to improve the goals and metrics that you have clearly defined. Understand if and how this technology can help you expand and optimize your existing methods. Furthermore, find how things relate and look at them from different angles. Enhance your abstraction capabilities, have an open mind, test assumptions, and identify biases. All of these “human” qualities are essential for successful AI implementations.
Some more resources
This will do for a morning read. But if you’re looking for a more hands-on approach, take a look at the short intro courses on Kaggle for Machine Learning and Deep Learning. Some Python knowledge is required to understand and finish them. You might also want to read about SQL (Structured Query Language) and how it is used when creating relational databases. Another great way to grasp how AI is making its way into modern digital marketing is by understanding and interacting with CRM and/or Marketing Cloud platforms by leading companies such as Salesforce and Adobe. Their tech stack is impressive and the possibilities with these tools and their AI engines are huge. Finally, check out this short YouTube video that really helped put things into perspective at the beginning.
Product Marketing @ MetaMask 🦊 || ex-Nike Brand ✔
Let’s connect: CryptoMailClub || LinkedIn || Instagram || Twitter
