Is AI the Cherry on Top of the Marketing Cake?

The following is republished with the permission of the Association of National Advertisers. Find this and similar articles on ANA Newsstand.

AI isn’t just a mysterious, ethereal machine swirling around every aspect of our daily lives, from physical spaces to cars to smartphones. It’s also an increasingly vital tool for brand marketers. AI isn’t just about making jobs seamless but enabling the right message to find the right person at precisely the right time — while also making the process of purchasing, learning, or engaging online and offline a lot easier.

AI is defined as “the theory and development of computer systems able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.” It’s an umbrella term that encompasses deep learning, machine learning, and natural language processing.

Although AI can accomplish a great deal, it’s not eclipsing human endeavor. In their book Human + Machine: Reimagining Work in the Age of AI, co-authors Paul R. Daugherty and H. James Wilson write: “AI systems are not wholesale replacing us; rather, they are amplifying our skills and collaborating with us to achieve productivity gains that have previously not been possible.”

While AI continues to evolve, it is sometimes misunderstood within the advertising and marketing industry. At root, AI is about personalization. It enables humans to live in a more connected world, with every touchpoint both trackable and understood. Armed with trackable and identifiable data regarding consumer behavior and preferences, marketers are able to understand different segments and demographics and tailor their products accordingly.

From smartphone manufacturers to data scientists, it’s easy to be confused about the fine details of AI — such as machine learning and deep learning — and how companies get a grip on best practices and uses. These days, nearly everything is powered by AI, like our voice searches asking Siri to find the best sushi restaurant in a specific neighborhood or Alexa to play a specific radio station.

Engagement, and how AI affects the way people communicate and function, is the Holy Grail for marketers, of course. AI is having a profound effect on consumer engagement, enhancing certain everyday tools and tasks while simultaneously giving people more information and the best options to educate themselves on products and services.

For instance, AI isn’t just about music preferences or shoe recommendations. It can be used for good. According to Wavemaker, the second largest media agency network in the world, The Washington Post has a “robot reporter” that has written approximately 850 short reports in the last few years using automated technology. The benefit is twofold: It eliminates tedious tasks for flesh-and-blood journalists so they can focus on writing more enterprising articles while also distributing important information to the Post’s audience.

In addition, Wavemaker reported that Johns Hopkins Hospital deployed deep learning to teach a machine how to diagnose early pancreatic cancer “based on recognizing textural changes on the exterior of the pancreas.”

But does AI go beyond providing menial tasks in order to focus on functions with more sophistication? Some of the evidence suggest it can, if directed by human intent, that is; AI is only as productive, innovative, and intelligent as the human programmers who create it. Time will tell. As of now, marketers must learn, track, analyze, and program AI in ways that will help distinguish their brand and sharpen their messaging architecture.

New Research

Future Delivery: AI and Email Marketing

Email marketing remains a vital cog in marketers’ wheelhouse. According to Loren McDonald, program director of marketing research at IBM Watson Marketing, by 2021 virtually every aspect of the email marketing process will be powered by some form of AI or machine learning. The technology will generate the greatest value in two specific areas: consumer innovation (creating better customer experiences through personalization) and workflow optimization (making lives easier and marketers more productive). Other machine learning applications include:

•     Subject line creation and testing
•     Predictive analytics
•     Content management and offer optimization
•     Reporting and productivity
•     Image and asset retrieval and tagging
•     Anomaly and struggle detection
•     Segmentation and personalization

So how do marketers get started with AI? First, they must define the problem that they want to solve, such as:

Struggle detection: Struggle detection recognizes breakdowns in the customer experience, and alerts marketers to take action. By understanding what people do when they go down a certain conversion path — and detecting anomalies — marketers can minimize waste.

Hyper-personalization and offers: Mercanto is an example of a platform that creates a unique combination of offers based on various consumer taste profiles and product attributes. The tool leverages natural language processing when receiving product information (such as color, descriptions, types of product, and user feedback and reviews) and finds patterns throughout words and terms in order to define the essence of each product. By observing consumers’ behavior — whether they’re buying, clicking, or searching a website — it builds a “taste profile” for each consumer.

Performance anomalies: Anomaly detection searches for patterns of online behavior that are either negative or positive. It can find correlations to possible causes, such as sending too many messages during a short period that led to a high bounce rate. It sees trends, spots any patterns out of the norm, and discovers what the root cause of the anomaly. If it’s wrong, it’s up to human supervision to improve its performance.

Content that outperforms human-generated content: Could a computer be better at writing email subject lines than a marketer? Phrasee is a tool and platform that leverages multiple aspects of AI. Several linguists interview clients, and the machine is fed that data, enabling it to capture each brand’s essence through phrasing, words, and tone. The machine provides 10 subject lines, which are then tested. Ninety-eight percent of the time, the winning subject line will be one of Phrasee’s (and not a human’s).

New customer insights: Contact fatigue, which occurs when too many emails are sent, is a severe problem that plagues most every email program. But help is on the way, thanks to machine learning. AI can gauge individual behavior and how consumers interact with a marketer’s messages, whether they opt to open, click, or unsubscribe. It learns how customers act as an aggregate (or individually), and identifies each customer’s individual tolerance to message frequency, scoring levels of fatigue to minimize opt-outs, and disengagement.

Key Takeaways

•     AI will change how marketers work: AI will change most jobs for the better, although it will be a challenging transition. A human will have to manage and train the machine. Ultimately, AI will create new, more efficient processes, and provide more time for strategy, creative, and marketing analysis.
•     Solve marketers’ data challenges: Marketing and email marketing are all about data. What you put into the machine will determine what comes out. AI will create a company’s single biggest competitive advantage, says IBM President-CEO Ginny Rometty.
•     Break down silos and merge creative with tech: AI will help marketers focus on the customer instead of different departments and break down corporate silos. In most companies, many employees do not know when other business units message to the same customers. Customer data platforms are a solution for structuring data in a single platform, with a 360-degree view.

Trending

A Different Voice for Marketers

Emerging voice technology enables users to make queries and receive responses via audio. The technology draws on content sources available online, which voice assistants interpret, feeding the spoken results via connected devices such as phones and smart speakers.

According to digital agency 360i and its client Roto-Rooter, voice technology presents marketers with a growing opportunity. Indeed, as of January 2019, Google Assistants were available on one billion devices, up from five million in May 2018, 360i says.

Voice-technology is an increasingly bigger aspect of people’s lives: 71 percent of owners of smart speakers, such as Amazon Echo and Google Home, use voice assistants at least daily while 44 percent access them several times a day. This trend stands to accelerate, as Juniper Research projects that the use of voice assistants in smart-home contexts will grow 1,000 percent by 2023.

The technology offers marketers multiple benefits:

•     Omnipresence: Voice technology promises to be accessible across all smart devices — smart TVs, speakers, phones, etc.
•     Trusted User Experience: The technology provides users with ease of use and marketers with another means of owning their messaging.
•     Adaptation: Adoption of the technology shows consumers that a brand is working to adapt to how they communicate, and thereby diminish friction in their interactions.

360i also reported that Amazon and Google are pushing for market-share dominance, thanks to the fact that their technologies are mostly open source, enabling billions of devices to incorporate them. 360i offered tips designed to help marketers capitalize on voice technology, including:

•     Recognize that marketers should be adding value to consumers’ lives through voice capabilities.
•     Ask voice assistants questions at scale to identify gaps or weaknesses in the available content, or white space in which marketers have the opportunity to satisfy an unmet need.
•     Recognize that voice technology requires a long-term strategy, involving intelligent deployments and ongoing promotions.
•     Take advantage of branded assets to encourage use of the company’s voice technology. Packaging, for instance, can be an effective means of educating consumers on how to use, or “invoke” a skill. Starbucks, for instance, encourages use of the technology on its cups, and 360i envisions a future in which Roto-Rooter uses the sides of its vans to alert consumers to their ability to access its DIY content through Google Assistant.

Digital Shift in Storytelling

Marketers are witnessing a massive technological shift — a new age of connectivity fueled by audio and visual media. Emotional storytelling has been a popular approach marketers have taken in the last several years to engage consumers, and technology only spurs the development.

Did you know that people in the U.S. spend four hours a day listening to audio? And younger generations spend seven hours patched in? Audio moments have become an integral part of everyday life, mobile being the main facilitator. Mobile essentially transports digital audio and enables audio, whether music or podcasts, to be part of people’s lives. This presents an entirely new way to reach people, whether they’re cooking in the kitchen or playing with their kids.

Pandora, an online streaming music service, shared how and why audio resonates and how contextually relevant ads drive receptivity during everyday moments, whether via relevant mindsets or personalized ad experiences.

Here are some stats on the current state of audio, per Pandora:

Pandora partnered with Nielsen to discover how people recalled brand ads through audio cues and identity versus recall of visual ads running on TV. The companies reported that audio increased brand and ad recall: 74 percent recalled relevant audio ads versus 65 percent of people who recalled relevant TV ads.

So what does this mean for marketers, particularly as they craft new campaigns? People are paying attention to audio in different ways than TV and people want to be “entertained” by TV versus wanting to learn and engage with audio. Relevancy equals receptivity, so contextually relevant audio ads boost ad recall.

Music, of course, is critical in setting (and influencing) people’s moods. To wit, 87 percent of responders said music improves mood while 86 percent of respondents said it relaxes them and 74 percent believe music motivates them. Sixty-four percent stated music helps them focus.

Pandora’s Music Genome Project provides a bit of road map for appealing to consumers. The service provides a truly personalized listening experience to users; each song has details collected on every track — 450 musical attributes altogether, including mood and other taxonomic factors that can help the algorithm suggest tracks based on category and relevance.

Knowing music powers people throughout their day caused Pandora to shift to a “mindset” mentality, or targeting consumers based on what they want and need through their preferences; personas are created through various moods, such as the type of music people like to listen to when they’re at work or play.

Personalization, of course, is a key strategy for all marketers: According to Pandora, 51 percent of marketers are increasing their investment in personalization. For audio-centric companies like Pandora, three facets drive personalization:

•     Dynamic audio: real-time creative messaging to listeners
•     Sequential audio: brand story told over time
•     Shorter-length audio: for short-term listening

These facets pull together audio, music, location targeting, and weather data in real time to create moments and ads that resonate and pertain to each user. Personalized audio ads increased lifts in purchase by 125 percent, and ad recall increased by 13 percent.

Best Practices

AI: The Big Picture for Marketers

Marketing and advertising executives weigh in on the state of AI, how to use the technology, and its limitations.

Michael Dobbs, SVP of Content Discovery at 360i: “Screens are really important to the experience. Using your TV, for instance, as a way to capitalize on the power of voice can create a really compelling user experience. Smartphones will also be a key part of creating that experience.”

John Zissimos, VP of Creative, Brand, and Media at Google Cloud: “Google Cloud’s biggest challenge was to make the ads, and Google Cloud, fun, simple, and accessible. Otherwise, what’s the point? Not everyone is a data scientist. The creative team, who works with our data team, had [to] make it fun. If it were just data scientists creating campaigns, the campaign and product would have never been made. We learned that you can’t do this by yourself. Partnerships are key to success. We also learned how to segment our audience better for this year.”

Jesse Wolfersberger, Chief Data Officer at Maritz Motivation Solutions: “It’s important to understand what predictions the technology is making for different groups. It’s not enough just to exclude demographic data because the training set can still influence the output in biased ways.

Transparency is the number one thing. You have to be transparent about what the data will be used for. But once it’s been used and turned into an algorithm, you also need to be transparent with users about why it’s making a particular recommendation to them. AI doesn’t really change the equation in a fundamental way. Data was already being collected and segmented; AI just allows that to be done with greater precision.”

Loren McDonald, Program Director of Marketing Research at IBM Watson Marketing: “AI cannot solve problems when there isn’t enough data. It cannot solve small problems that aren’t scalable. One challenge with tools like Phrasee and such is that for the tool to be confident in its recommendations, it needs millions of interactions and customers. The problem has to be really big, with enough data to solve it. It’s really easy to build self-driving cars that drive down the freeway, with lanes, in the middle of the day. That is a relative piece of cake. An autonomous car in Wisconsin in the winter with snow and no lanes — that hasn’t been solved yet. The creative part is also a way away; the computers don’t understand creative problems. Start by looking for things where there are simple, understood changes in behavior and patterns, which can easilybe uncovered and fixed.”

Case Studies

Google Leverages AI and Machine Learning

Data-driven creativity merges art and science. Google did it in 2018 when it ran its NCAA ads during the Final Four matchups on the cable station TBS — the first real-time TV spots featuring AI predictions. Ads for Google Cloud ran before the Kansas-Villanova and Michigan-Loyola Chicago basketball games, with a second set of ads running during halftime of each game. The point of the ads? Predicting the final scores. The ads, which were created with agency Eleven, didn’t actually end up predicting the scores, but they did make for compelling content. Using data to predict outcomes in ways that are fun and conversational is not only an interesting concept. Google hopes the practice will encourage businesses and nonprofits to use Google Cloud as a way of using large amounts of data for innovative projects that integrate the way people live, work, and interact with technology. Google shared tips on how the company succeeded with its campaign:

•     Respect the user and customer. Google Cloud’s focus is all about respecting the user while celebrating individuality. In order to do this, the focus must be on the customer journey, realizing and anticipating the needs of both individual users and businesses.
•     Find an opportunity. Opportunities rarely just present themselves, so Google recommended finding ways and situations to be innovative with data. Google Cloud, for instance, helped people experience a moment together by leveraging a media campfire: March Madness. The company demonstrated how Google Cloud helped the NCAA glean intelligence from its data and give consumers something valuable about basketball that they can’t get anywhere else. In this case, it allowed fans to be more engaged with the game and, for the NCAA, illustrated a close connection to its players and games.

Google Cloud was also able to generate insights regarding its audience:

Businesses are sitting on mountains of untapped data, and the C-suite knows this data holds the key to market advantage. Yet only one percent of all business data is being mined for the vast potential knowledge it holds. This is when Google Cloud came up with a new slogan: “Know what your data knows.”

To harness the data, Google Cloud created the first real-time TV spots featuring AI predictions. The company analyzed the first half of both the “Final Four” and championship games and then pumped the data into models; the second-half prediction was created through a custom-built Google Cloud software. Within 10 minutes, the ad was rendered and subsequently aired.

Businesses are sitting on mountains of untapped data, and the C-suite knows this data holds the keys to market advantage.

To accomplish this, 279 unique animations were created for situations like rebounds and shots. These animations were then uploaded to an app called Stardust, which cataloged all of the permutations that might happen. The challenge? It had to be done at a speed that had never done before.

•     Partner up. Partnering with the NCAA allowed game data to be accessible within Google Cloud. This produced a vertitable toolkit of machine learning, with a 70 percent accuracy rate. In the end, Google Cloud racked up the following: 91 percent lift in product interest; 30 percent increase in new site visits; 19 percent increase in time spent on site; 13 percent boost in unaided brand awareness; 11 percent rise in consideration, and a 42 percent climb in brand-search volume.

How HSBC Generates Loyalty with AI

Maritz Motivation Solutions found a way to apply AI on behalf of the credit card industry through its work with HSBC’s Loyalty Program. HSBC found that its cardholders were resistant to using the points they accumulated through the company’s rewards program.

The banker attributed this to the unfamiliarity of the program’s rules and the fact that navigating them required the use of math. HSBC sought to solve this problem by using AI to serve users personalized recommendations for how to redeem their points. The technology drew on disparate data sources, including purchase and redemption histories, to identify the rewards most likely to satisfy both the rules of the program and a given consumer’s preferences.The initiative was so effective that more than 70 percent of users redeemed their points for the item that AI suggested.

Maritz touted customers’ evident satisfaction with the work of AI justify using the technology to offer another selling point that credit card companies could use to differentiate themselves: “intelligence,” meaning the intelligence absorbed to understand a given consumer and to capitalize on that understanding to help him or her make financial decisions and plan for the future.

Spotify Tunes into Sharper Personalization

With 248 million-plus users logging in, Spotify has access to oceans of user data, which swell to even larger sizes when subjected to thoughtful and machine-aided analysis. Spotify, and similar music-streaming services, knows not just when its users are logged in, but on which devices. Moreover, based on what users are listening to, it can infer what people are doing. For instance, when Jane Doe is listening to her “Pilates” playlist, the service knows not just that she is working out, but the specific kind of exercise she is performing This information can help guide Spotify in its efforts to introduce users to new music that suits their mood.

Spotify can capitalize on this data to understand the type of content to offer to users initially. The specific music that users listen to is, of course, a source of data unto itself. Its attributes can be especially revealing when analyzed by AI to generate “machine-powered personalization.” Spotify identified three types of attributes to harness the effort via AI:

•     Song attributes, such as pitch, key, time signature, rhythm, and tatum (the smallest cognitively meaningful subdivision of the main beat of a piece of music).
•     Acoustic-semantic attributes, e.g., the playlist title cross-referenced with sound attributes such as “danceability” and “energy.”
•     Genre, not just “macro-genres” such as rap and country, but the 37,000 microgenres identified by Spotify, ranging from “progressive house” to “vapor trap.”

Sean Kegelman, global head of data and audience solutions at Spotify, talked about how the company has used a client’s first-party data to build value for that person. “We tend to be able to help partners a lot when they can give us DMP or CRM data,” he says. That helps to “illuminate things like consumer ideas, which can shape a campaign
or more broadly a brand strategy — for instance, how to select the right background music when communicating with a discrete audience or segment, or how to do sponsorships or event activations.”

Personalization is a critical component of how Spotify deploys AI, Kegelman adds. “Our level of personalization discourages a lot of sharing,” he says. “You don’t want to see music recommendations tailored to a friend’s taste on your service. However, we also offer plans like family plans that encourage group use while preserving distinct profiles with personalization for individual users.”

Source Information

“Shopper Marketing: The Good, the Bad, and the Ugly.” John Bajorek, EVP of Strategic Growth and Innovation at WD Partners. ANA Shopper/Commerce Marketing Committee Meeting, 3/20/19.

“The New Age of Digital Storytelling.” Brendan O’Marra, Director of Digital Marketing North America at BIC; Catherine East, Managing Director at Doner CX; Kendra Tal, Director of Ad Innovation at Pandora; Susan Cylenica, Ad Innovation Lead at Pandora; Joe Prota, Manager, Global Social Practice Lead at IBM. ANA Digital and Social Committee, Southeast Chapter, 9/10/19; ANA Digital and Social Committee, 9/26/19.

“Voice Technology for Marketers.” Michael Dobbs, SVP of Content Discovery at 360i; Sally Bayer, VP of Marketing at Roto-Rooter. 2019 ANA Media Conference, 4/10/19.

“Machine Learning and AI for Marketing and Other Disciplines.” Devyani Sadh, CEO and Chief Data Officer at Data Square; Leo Kluger, Principal Data Scientist at IBM; Matt Ryan, Director of Database Marketing at St. Jude Children’s Research Hospital; Jim Griffin, Americas Director at Cartesian DataSciences; John Friedmann, Manager of Analytics at Deloitte; Amit Deshpande, SVP of Analytics at Epsilon. ANA Analytics & Data Science Committee Meeting, 2/27/18.

“How Google Used AI and Machine Learning in a Brand Campaign.” John Zissimos, VP of Creative, Brand, and Media at Google Cloud. ANA Brand Management, West Coast Chapter Committee, 1/24/19.

“How HSBC Generates Loyalty with AI.” Jesse Wolfersberger, Chief Data Officer at Maritz Motivation Solutions. ANA Relationship Marketing 1-Day Conference, 4/2/19.

“How AI Will Affect the Future of Email Marketing.” Loren McDonald, Program Director of Marketing Research at IBM Watson Marketing. 2019 ANA Email Evolution Conference, 4/26/19.

“How Spotify Uses AI.” Sean Kegelman, Global Head of Data and Audience Solutions at Spotify. ANA Digital Marketing and Innovation Conference, 11/15/19.

“Digitas: Exploring AI-Born Creativity.” Adam Buhler, SVP and Experiential Lead at Digitas. Marketing Futures Committee Meeting, 5/9/19.

“Making AI Work for You.” Wavemaker, 2019.

“Futures Forward Studio: How AI-Enabled Technology Transforms the Personalization of Retail.” Greg Paull, Co-founder of and Principal at R3; Tina Gaffney, Head of Customer Success at Realeyes; Alice Chen, Senior Brand Manager at Hotels.com; Greg Pal, Chief Business Officer at Automat. 2019 ANA Brand Activation Conference, 5/15/19.
Source

“Is AI the Cherry on Top of the Marketing Cake?” Insight Brief written by Joanna Valente, Manager, Content Strategy, Marketing Knowledge Center, ANA. Designer: Amy Zeng, Marketing and Communications, ANA. Editor: Matthew Schwartz, Senior Manager of Marketing Communications, ANA. © Copyright 2019 by the Association of National Advertisers, Inc. All rights reserved.