Are You Really Smart About How AI Works in Marketing?

In its widely talked about State of Marketing Report, Salesforce reports that just over half (51%) of marketers are using AI in one form or another, while another quarter plan to test it over the next two years.

A smaller study of over 500 search, content, and digital marketers by BrightEdge found that just 4% have implemented AI (that’s not a typo).

Who’s right? Salesforce, which reports one in two marketers is using AI, or BrightEdge, which puts the number at one in 25?

The answer may be “neither.” That’s because many marketers (and business leaders as a whole) are confused about which technologies are genuinely AI-powered and which simply rely on advanced algorithms and analytics.

As Luis Perez-Breva, head of MIT’s Innovation Teams Program and research scientist at MIT School of Engineering, explains, “Most of what the retail industry refers to as artificial intelligence isn’t AI.” He says many “confuse analyzing large amounts of data and profiling customers for artificial intelligence. Throwing data at machines doesn’t make machines (or anyone) smarter.”

Rather, AI’s promise is what is often called relevance at scale. It’s the ability of machines to crunch massive datasets and data lakes – structured and unstructured data – and optimize decision-making in a way that algorithm-enabled humans cannot achieve. Perhaps most importantly, in an AI-enabled system the machine learns and improves without human input.

Rather than ask, “How many marketers are using AI?,” the more apt question may be, “What are you doing with it?” Let’s examine some of the ways companies are using AI-led initiatives to make the most of AI’s promise.

Using AI for personalization

Marketers have long practiced personalization in content marketing, developing over time more sophisticated ways of personalizing the customer journey – whether through marketing automation and progressive profiling or using programmatic advertising to support our content path. The idea is that as we learn more about our customer or prospect and fill in information about that person’s needs, budgets, and interests, we can create unique, personalized experiences that educate and delight the person.

Now we are entering the era of hyper-personalization: the ability to personalize not just by persona, profile, or the trail of breadcrumbs people leave on your site, but by a massive set of user details and signals, analyzed and made actionable by machines.

The retail industry is the most talked about application of AI-led personalization, but most examples you read about don’t really fit the definition of AI … they’re just really good personalization.

The examples that seem to cross over – from algorithm-driven personalization to AI-driven personalization – are those in which the AI sifts through data from multiple channels and sources, learning which signals matter in which circumstances and evolving its approach over time. The key variables that influence how one customer interacts with your brand may be completely different from the variables that define another, multiplied millions of times across each person, each channel, and each step of the process – and changing constantly.

Using AI for voice-searchable entertainment and education

A less common but exciting application for AI-enriched content? Virtual assistants. Alexa (Amazon) offers developers the chance to build “skills” on its platform. Alexa Skills help customers answer questions, gather information, and even control internet-enabled devices and appliances. (To be fair, there’s disagreement about whether Alexa is an AI technology or just an advanced natural language technology – another nod to the problem of assessing AI adoption.)

Companies far and wide are racing to launch Alexa Skills – both to inform and delight customers as well as to test out the channel’s promise.

Entertainment

Content-rich brands are delivering entertainment and information via Alexa Skills. Disney’s Character of the Day Skill introduces a new character each day from Disney, Pixar, Marvel, and Star Wars. Or you could try out Cat Translator to understand the “why” behind weird cat behavior.

Real-time news

Media companies have been among the first to offer content snippets via Alexa Skills. If you enable the NPR News Hour Skill, for example, you’ll have access to a five-minute news summary, refreshed every hour. Big brands are quickly jumping in too. J.P. Morgan customers can access investment news: “Send me the latest research report from Joyce Chang” or “Send me the tear sheet for eBay.”

Customer service and engagement

Global consumer brands are enabling e-commerce, customer service, and analytics using Alexa Skills. The Capital One Skill lets you ask Alexa, “How much did I spend at Target last month?” or “When is my mortgage payment due?”

For content marketers, there are interesting opportunities to deliver education and entertainment via voice-enabled search. Beauty brand Wunder2 was the first in its segment to launch an Amazon Alexa Skill. The company offers a daily beauty tip via Skills, from how to thicken the appearance of your brows to how to achieve healthier looking hair. As one reviewer explained, “It’s very cool when I can get the latest beauty tips while having my hands free to apply my makeup.”

Wunder2 co-founder and CEO Michael Malinsky tells Forbes, “As a business, we are fascinated with the rapid integration of AI into people’s lives. We think the level of adoption will exceed many people’s expectation and create fluid recommendation experiences using AI technology found in Google Home, Alexa, and the recently launched Apple HomePod. It is something we are absolutely developing already.”

Using AI to put email on steroids

For marketers, AI-enabled decision-making for customizing and delivering email (i.e., dynamic emails) could be a game-changer.

Once upon a time, marketers would ask, “What’s the best time of day to send out our email newsletter?” Through trial and error, marketers discovered that certain days and times yielded higher open rates on average.

AI, however, allows marketers to send emails based on the open histories of individual users (or people like him/her in the absence of better data). And no longer will marketers send promotions to huge swaths of their audience. Instead, promotions will be designed uniquely for prospects based on a wide range of signals, from cart abandonment in retail to which times of day an individual is most likely to sign up for a conference. Finally, AI will enable much more customized and nuanced customer journeys. That leads to our next AI application – one which is too often misunderstood.

Using AI to write

Long decried as evidence that AI will herald in a new soulless age, machine-made content is one of the most controversial applications of AI … but, under the right circumstances, it may be the most pro-creative. Let me explain.

As machine-made content becomes better at approximating human language, there’s a clear case for its use in content marketing. Not all content generated by marketing needs to be highly creative and witty, after all. Many organizations are already using machine-generated content, such as Edmunds generating vehicle profiles based on manufacturer data and Homesnap publishing community profiles based on publicly available data. The best applications are those in which there’s a need to publish at scale and the content is somewhat “modular” or easily put together from pieces and parts.

And, if you’re not convinced, perhaps this will change your tune. Even The Washington Post uses machine-generated content. According to Digiday, as of September 2017, the paper’s robot writer (a solution from Heliograph) had published 850 articles and tweets like this one:

Landon beat Whitman 34-0; https://t.co/V6zVPi7a9O @LandonSports @koachkuhn

— WashPost HS Sports (@WashPostHS) September 2, 2017

The key is in how you pair the robot to the writing. For The Washington Post, Heliograph generated articles about local political races, where the paper didn’t have the resources to assign reporters but had data to fill in the story. It also published short summaries about the Olympics in Rio via machine. (The paper reports that four employees previously took 25 hours to collect, analyze, and report on a small portion of local election results. Using Heliograph, The Washington Post created more than 500 articles generating 500,000 views.)

And therein lies the most powerful promise of AI: to release marketers from the mundane to focus on more creative and fulfilling efforts. Marvin Chow, vice president of global marketing at Google, writes that artificial intelligence and machine learning “will spark new ideas and push the boundaries of creativity. With new tools, what will makers, artists, and musicians design? And how will that affect the marketing world we work in?” The full vision is still out of reach, but early signs point to a machine-led period of creative efficiency.

A version of this article originally appeared in the August issue of  Chief Content Officer. Sign up today to receive your free CCO subscription.

Discover more about how to use AI (and how not to use it) at Content Marketing World Sept. 4-7 in Cleveland, Ohio. Register today and use code BLOG100 to save $100.

Cover image by Joseph Kalinowski/Content Marketing Institute