Get smart quickly about machine learning & marketing

5 practical tips for becoming a machine learning marketing ninja

The increasing prominence of machine learning in marketing is changing the way the industry operates. No doubt, this evolution is exciting, as it opens up new opportunities to connect with customers in ways that weren’t possible just a few years ago. But it is also terrifying, as the race between the data haves and have-nots intensifies, and marketing FOMO sets in. The good news is that while it may feel like everyone else has already become an AI marketing ninja, the truth is that the industry as a whole is still learning. According to a report by Salesforce, only 26% of business leaders have confidence in their organization’s ability to develop an AI strategy. And this statistic aligns with my own experience working with marketing clients across a range of industries: most companies are at the beginning stages of using data and machine learning in their marketing (but they are learning quickly).

The challenge for marketers right now is that the appetite for AI-infused marketing solutions is high, but the understanding of what AI actually does, is low. Unfortunately, this breeds an environment where vendors play fast-and-loose with the term “AI” and marketers are at-risk of being sold digital snake oil. As an engineer, this dynamic frustrates me. So, I’ve outlined five practical steps for getting smart about marketing with machine learning quickly, in an effort to help marketers become more informed buyers, whether they are evaluating AdTech solutions, interviewing agencies or building their own in-house data science teams.

1. Define your marketing objectives

Before you start exploring the possibilities of infusing machine learning into your marketing, first get crystal clear on your marketing objectives. Machine learning is not an end in itself; it is a means to an end, to help you achieve your marketing goals. If your goals are clear, you’ll have a grounded set of criteria with which to evaluate machine learning technologies. For example, if your goal is awareness, and your objective is to reduce your cost per acquisition by 15% so that it is lower than your average customer life value (CLTV), then you should be looking for solutions that use predictive models to increase conversation rate and/or reduce the cost per impression or click. Alternatively, you might look for solutions that use personalization and smart recommendations to increase the amount each customer spends, thereby increasing CLTV. Both strategies use machine learning to improve your marketing ROI.

The thing to understand is that, despite all the hype, machine learning is not about making a general-purpose AI that can replace your marketing intern. Rather, it is a class of specialized algorithms designed to optimize highly specific tasks using data. Machine learning does not change the fundamentals of marketing. Like all AdTech/MarTech, you need to assess whether a machine learning solution is likely to increase your marketing ROI after taking into account to cost of the solution itself.

2. Gain a basic understanding of how machine learning works

Machine learning is such a different approach to solving problems that it’s important for all business leaders, especially marketers, to have a basic understanding of how it works under-the-hood. If you don’t have a conceptual mental model for how machine learning functions, then all purported AI solutions will look like magic and you’ll get lost chasing shiny objects that don’t perform.

So let’s start by defining the term properly: Machine learning is a subset of the broader field of AI that focuses on a category of specialized algorithms designed to optimize a specific objective (e.g. maximize clicks or identify fraudulent transactions) based on historical and/or real-time data. There are other fields in AI beyond machine learning (e.g. Expert Systems), but in recent years, machine learning has dominated, so for all practical purposes, the term “machine learning” and “artificial intelligence” have become synonymous. The real difference is that data scientists and engineers prefer the term “machine learning” while salespeople and futurists use the term “AI.”

My two favorite recommendations for machine learning newbies are a series of videos, one produced by Microsoft and the other by Google. These both provide a gentle introduction to machine learning before diving into some of the tactical details that illustrate how machine learning algorithms work. Another great option is taking on online course geared toward beginners. Udacity recently launched an Intro to Artificial Intelligence course taught by two AI heavy-weights, Sebastian Thrun (Udacity’s founder) and Peter Norvig (Director of Research at Google). If you have the time and inclination, this is the best way to get up-to-speed.

3. Gain an understand what machine learning is NOT

There are a lot of technologies out there masquerading as machine learning. As an informed buyer, you should keep an eye out for the technology buzzwords that are often conflated with AI. Here are a couple to watch out for:

  • Data-Driven Marketing: The use of data in marketing does not, in and of itself, equate to machine learning. For example, you might use first-party customer data to categorize website visitors into one of a handful of segments, and then display different content to users in each segment. This is data-driven marketing, but it is not machine learning. Yes, you are using data, and yes, you are doing personalization, but the rules for segmenting and displaying content are created manually by a human. A machine learning approach, in contrast, might segment users automatically based on historical data using a clustering algorithm like K-Means, or try to predict the content that a user is most likely to engage with using a method like Contextual Multi-Arm Bandits.

  • Big-Data: Many machine learning algorithms, especially neural networks, require vast amounts of data to learn. As a result, big data (i.e. large sets of data that can’t be stored or processed by a single server) is often associated with machine learning. However, big data does not equal machine learning. I’ve been on big data projects where the end goal is to provide traditional descriptive analytics (i.e. summary data and dashboards), and I’ve also been on machine learning projects which didn’t require big data to generate predictions. The two are related, but not synonymous.

It is important to understand machine learning in enough detail to be able to distinguish ML solutions from non-ML solutions, but it is even more important to remember that machine learning is the means, not the end. If you discover a promising MarTech solution that you think will help you achieve your marketing objectives, then who really cares if it’s using AI or plain old data-driven marketing, as long as it gets job done?

4. Ask questions about the technology

When evaluating a potential machine learning solution, don’t be afraid to ask a lot of questions about the underlying technology. This serves two purposes: 1) it allows you to get a feel for whether or not the vendor is full of hot air, and 2) it is a great way to learn about machine learning for free! Here a few questions to ask:

  • Have you published any academic papers? Many early-stage machine learning startups are spun out of academia and the technology is based on the founders’ research. Legitimate data science innovators publish papers on their work. If the vendor claims to have a “break-through algorithm” that serves as the basis for their competitive advantage, ask to see the research. You might not understand all of it, but it will give you sense if it’s real or not.

  • What machine learning algorithms do you use? By definition, a machine learning solution is based on a machine learning algorithm, so the best way to assess if it’s legit to ask which algorithms the company is using. You might think this is proprietary information, but most people are willing to share it because ML algorithms are really a family of algorithms, each with an enormous number of variations and optimizations. So by asking what algorithms the product utilizes, you’re really asking what their general approach to machine learning is.

  • How much of a performance lift is typical among your customers? At some level, this is the most important question. After all the magic AI fairy dust has settled, you want to know how the solution will improve your marketing ROI. Since the field of machine learning is singularly focused on optimization, you should expect a good vendor to be able to explain this question clearly.

5. Understand where the creative fits in

Machine learning and data can do many powerful things but producing compelling marketing creative is not one of them. Humans still need to be in loop for ideation and creative production, and depending on your objectives (e.g. brand building vs. demand gen), they may need to be involved a lot. Machine learning vendors tend to be technology-focused, so they often downplay the importance of creative. But bits and bytes don’t provoke a change in customer behavior—great creative experiences do. So, it is critical for marketers, as stewards of the brand, to assess how the creative process will be impacted by machine learning. Will all creative be optimized for the lowest common denominator? Or will be there be a place for big, bold ideas to shine? How will creative by QA’d if there are hundreds or thousands of variations? How will learnings from the data make it back to the creative team as feedback? All these questions will need to be addressed by the marketer evaluating machine learning solutions.


You don’t have to be a data scientist to be effective at using machine learning to improve your marketing, but you do have to arm yourself with some basic information. Fortunately, when it comes to assessing machine learning solutions, a little knowledge goes a long way. By asking a lot of questions, reading up on machine learning, and adopting a healthy degree of skepticism, you’ll become an AI marketing ninja in no time.