Artificial intelligence—you've heard the term, you know it's a trend, you know you're "supposed" to be using it, but if you were asked to explain what it means for marketers in a few short sentences, could you do it? If not, then you've come to the right place.
A quick Google search of "marketing AI" returns 950,000,000 results. And although Google does its best to surface the most relevant content to the first page (using AI to inventory, categorize and label, and recommend content, might I add), I doubt you have time to sift through each mention until you find a piece that actually gets down to brass tacks.
As a marketer, the things you likely want to know are these:
- What new, AI-powered marketing technology should I spend budget on?
- What time investment are we talking about in relation to implementing AI?
- Which marketing functions are best suited for AI? Which should I just leave alone?
AI sometimes feels like "the man behind the curtain"—elusive, complex, and a little scary. Instead of skimming by with surface-level knowledge, marketers should learn more. So here are a few questions, definitions, and tactics for evaluating marketing technology solutions that claim to be "powered by AI."
Breaking Down the Buzzwords
Artificial intelligence (AKA intelligent automation)
My all-time favorite definition of AI for marketers (and there are many definitions) comes from Paul Roetzer of the Marketing AI Institute: "AI is technology that automates a task previously done by a person." Pretty simple, right?
Every time you see AI in the context of martech, just substitute out the term "artificial intelligence" for "intelligent automation," which can mean one of two things in marketing:
- Recommending. Some marketing software predicts which action will have the most positive outcome in order to recommend a next step in a series of events. Think of these small recommendations as stepping stones on the way to fully automating a given task. A few examples include offering content topics for a blog post, or suggesting email subject lines.
- Automating. Automating builds on recommending. To qualify for being automated, a task needs to be routine and repeatable; the goal needs to be specific; and the steps to achieve that goal must follow an exact set of rules. Tasks that fall under this umbrella include running programmatic ad campaigns and triggering the next email in a journey-driven series.
As we collect more and more data, and the capabilities of marketing tech improve, the tasks we're able to automate within marketing will certainly increase.
Should you be worried that all functions of marketing will be completely automated? That's simply not the case yet. The likely evolution will be that the more functions are automated the more opportunity for marketing strategy and creativity.
Data science (AKA the tech that enables AI)
When you think about science, the first few things that likely come to mind are test tubes, beakers, and your 9th grade biology class—not data. But if you ask anyone who actually builds marketing AI for a living, you'll get a different answer.
According to the Berkeley School of Information, data science is the practice of "organizing and analyzing massive amounts of data," and to be an effective data scientist, one must be "able to identify relevant questions, collect data from a multitude of different data sources, organize the information, translate results into solutions, and communicate their findings in a way that positively affects business decisions."
The three most important pieces of data science to understand are as follows:
- The data source. A consistent, repeatable process that can be measured is referred to as the data source. It acts as the information input. An example of a data source is the open rate on an email. If recommending email subject lines, a data scientist could look at the open rate of previous emails, determine which had the highest open rates, and suggest a subject line based on that data.
- Big Data. Access to a large number of observations from a specific process is called "Big Data." For example, credit card companies use transaction records from a variety of customers to power fraud detection technology. Which transactions are common, which seem out of place as compared to the norm?
- Machine-learning. Organizing and analyzing structured data (like website visits and purchase data) or unstructured data (like images or written content), and making predictions based on that data, is known as machine-learning—an umbrella term that is applied to a range of data science methods. An example of machine-learning is scanning images and tagging them in a searchable database based on identified objects. Think about machine-learning as a set of techniques using machines (i.e., computers) to help us (i.e., humans) learn something specific.
Questions to Ask When Considering AI-Powered Marketing Tools
Next time you come across a too-good-to-be-true piece of marketing tech that promises to exceed your KPIs and make your boss smile, ask yourself the following:
- Which marketing task is being intelligently automated? Do I need to automate that task?
- Does the tech come with its own Big Data source, or will I need to provide it? Do I have the appropriate amount of data? Will I be able to connect my data source to the tech, if needed?
- Is there evidence of the technology either making good recommendations for or automating one of my tasks?
When considering new technology, it's all about knowing the right questions to ask. So, hopefully, those questions will help make AI a little more approachable for your marketing team.
* * *
Artificial intelligence, Big Data, and machine-learning aren't going anywhere, so the quicker you are able to determine the need-to-know info about the systems you are considering implementing, the faster you will be on your way to staying ahead of marketing's biggest disruption.
