square peg being forced into a round hole
PHOTO: shutterstock

A writeup from eConsultancy lists 15 different categories of AI in Marketing — yikes! From best offer predictions to social listening to smart segmentation to facial recognition, vendors have sprinkled AI across nearly every stage of the customer lifecycle. And in many cases, marketers are working backwards, getting sold on new AI offerings that may carry risks and may not be the best fit or provide the biggest improvement in performance. 

Gartner analyst Andrew Frank recently wrote, “AI’s capacity to transform marketing is obscured by a fog of hype, but the breakthroughs are real. Marketing technology leaders need to engage in AI initiatives or risk being blindsided by disruptive AI-enabled competition.” 

So how does a marketer cut through the fog of hype to get to the good stuff? BI Intelligence found that about half of all marketers are using some form of AI. For such a new technology, these adoption numbers are high. It’s likely many marketers have grabbed at low hanging fruit from current partners, rather than taking a considered approach. 

When Considering AI for Marketing, Learn What’s Possible 

It's better for marketers to take the lead and set up an AI reconnaissance mission, rather than passively reacting to pitches from current and prospecting vendors. Marketers should allow for an information gathering phase without any preconceived notions of what direction will eventually be taken, and without any promises to executives or partners. There are a lot of things to consider in addition to what cool thing the AI offering can do or the eventual ROI. One way to start is to make a list of the elements of the customer experience or marketing process that could use a boost and then search for solutions.

AI has popped up in many corners of marketing that could be immensely valuable for a brand — if they spend the time to find them. The key is to understand what output the technology will provide. It’s not a quest to find a technology, but rather a quest to determine the best possible output you’ll get from that technology.

Leading marketers are implementing AI in very smart ways to create valuable outputs, even if they aren’t particularly sexy. Zappos implemented AI to improve the context of their searches. With so many products, it’s important for people to get what they want quickly. Netflix uses AI to personalize every image they present — down to the thumbnails for a show — for a tailored content experience, knowing that tailored content is a key to its customer loyalty.

There is a lot to understand about an AI offering before knowing if the output is worth the investment. For example, some brands will be comfortable with black box algorithms, while others want more transparency. Marketers must understand what data inputs, scale and technical requirements each offering requires in order to function. Some offerings are unproven and at a very early stage, while others are robust. Some AI requires a lot of customization, some works out-of-the-box. And of course, it’s important to deeply understand the effect on customer experience — and the known and unknown risks.

Related Article: Where Does AI Fit in Your Marketing Strategy?

Next, Evaluate AI Options Against Your Marketing Strategy 

Gathering so much data changes the evaluation stage significantly, and gives marketers the insights they need to evaluate the output they can expect. Rather than being caught up by a cool demo or powerpoint deck, marketers have the insights to know how long it will take to implement, what kind of scale a test needs to be, and how relevant the outcome will be to their own strategy.  

Take a luxury brand that wants to increase engagement with loyal customers both online and in stores by making their brand more relevant. By evaluating the entire landscape, they are able to compare and contrast different paths they could take and the pros and cons each one provides. Compare this to the alternative, where they jump on a cool AI offering, and then marketing and IT get swamped in the hairy details as they work to implement it, wishing they had asked the hard questions up front. Or worse, a new technology is implemented and ends up freaking out customers rather than creating more relevance.

With intelligence about different AI offerings, marketers can narrow the options to those that are actually feasible to implement. From this smaller list, it’s possible to run scenarios and compare the output, user experience and potential returns, as well as the company’s appetite for unknown outcomes. That same luxury brand could now confidently compare the costs, time to implement and CX effect of having facial recognition at kiosks in store vs. implementing a virtual assistant on their website that uses offer and pricing prediction.

Related Article: How Bringing Machine Learning Into Marketing Improves Business Results

Use Outcomes to Select the Right Approach

AI is very powerful, and we’ve all been subject to its effects. It’s important for marketers to move forward thoughtfully. Marketing AI today can automate and improve productivity, relevance and customer experience, but there are many unknowns. Consumers have publicly trashed malfunctioning chatbots and terrible recommendation engines. Algorithms could go haywire when an overnight data feed goes down. Customers could simply dislike the new experience. 

Luckily, with a research-based approach, marketers who have done their homework can be confident they’re thinking about the big picture. They will know how much data they would need, how valuable different options are to their customers, and how to avoid unwanted outcomes. This makes it easier to say “no” to projects that sound sexy, but simply don’t fit with the marketing strategy, and instead follow a path to sustained success.