Artificial Intelligence and Digital Marketing

Artificial Intelligence has been complementing traditional marketing techniques for almost 30 years now. Typical mediums for traditional marketing are newspapers, radio stations, and commercials. Digital marketing on the other hand only needs the web to display a vast amount of different advertisements. While the mediums of communication may differ greatly, the most important difference is that digital marketing leverages concepts such as machine learning, neural networks, and other technologies to develop its own algorithms. Digital marketing wasn’t always as involved as it is now and has notably changed throughout the 3 major iterations of the internet: Web 1.0, 2.0, and 3.0. All of these different iterations have brought more functionality to users which has also allowed companies to think outside the box when it comes to interaction with their consumers. During recent years we have experienced an uptick in the amount of industries and major corporations utilizing and researching AI in innovative ways. Inevitably, this rise in applications has caused a surge of questioning in regards to the ethics, purpose, and litigation surrounding Artificial Intelligence and Marketing.

Illustrates projected growth of the digital marketing industry from 2019-2028.

History

Web 1.0

Web 1.0 brought us a very basic version of the internet. Coined in 1994 by Tim Berners-Lee, we saw people use the internet to search and purchase products or services; however, they were not able to contribute their own ideas to the web. During this time, the web was sprawled with many personal web pages that, by today’s standards, were very basic and consisted of mainly static pages. Companies attempting to go digital would need to get around advertisements being banned while surfing the web.

Web 2.0

A big upgrade from Web 1.0, Web 2.0 didn’t actually change any of the technical specifications that came with its predecessor. Instead, Web 2.0 aimed to alter the way that web pages were designed and interacted with. We saw a plethora of new technologies emerge that are still used today, such as JavaScript, Ajax, and CSS. This resulted in the first interactive and collaborative websites being created, inevitably bringing us the birth of social media. With the average person now being able to interact with friends, family, and strangers across the globe, companies wanted to be able to interact with their customers. Corporations began to realize that since the Web became interactive, social networking website usage has increased immensely. To get a wider net of customers, companies started to spend larger percentages of their budget on digital advertising and marketing. This resulted in digital advertising and marketing becoming a $2.9 billion industry by 2004.

Web 3.0

Often referred to as the Semantic Web, Web 3.0 was an extension of the existing Web. Instead of having a front-end focus like Web 2.0, this upgrade was more back-end oriented. In other words, we were able to make our machines much smarter. The Semantic Web provided an efficient way to represent data on the Web, or in a database, to a machine. Computers were, and still are, able to extract meaning from the data utilizing relationships, rules, and logic. With the new functionality implementation of technologies such as Artificial Intelligence and 3D Graphics were possible. We can also access content across multiple services being used across the globe, a concept known as ubiquity.

Growth of the Digital Marketing Industry

Knowledge about the history of the Web is critical in understanding the recent growth of the digital marketing industry. Transitions from Web 1.0 to Web 2.0 brought us interactive ads. Transitioning from Web 2.0 to Web 3.0 has presented us with behavioral advertising, allowing companies to target different audiences in a plethora of ways. Because of these developments in the Web, we have seen tremendous growth throughout the industry in the past few years. It was estimated that the digital marketing industry was worth $350 billion in 2020 and is projected to grow to $786.2 billion by 2026. Since the pandemic, there have been understandable budget cuts which have brought down the overall digital marketing spending; however, that does not mean that the industry’s value is shrinking. With the increase of people working from home, there is now an even wider range of people for companies to advertise to. Companies are also in a better position to start increasing their digital marketing budgets as we continue approaching a post-pandemic world.

AI Integration into Digital Marketing

Since the creation of Web 3.0, AI development has been at the forefront of giving each consumer a unique Web experience. Not only is this beneficial to the end user, but companies are seeing massive profits by digitizing and automating their advertising network. The McKinsey Global Institute states that AI and Machine Learning are on track to generate $1.4-$2.6 Trillion in value over the next 2 years. With the digital marketing industry doing well, it seems that AI is the bottleneck between the two. When asked about the state of the marketing industry, 70% of respondents reported that a lack of AI training and education is holding the marketers and industry back. Job openings for artificial intelligence engineers are only increasing as time goes on. Burning Glass found in 2020 that over the next decade projects involving AI will grow by 40.1%. When job demand for a skill increases, educational institutions that can effectively teach that skill will become highly sought after. The gap between AI and the digital marketing industry should narrow over time as more qualified individuals enter the workforce.

Cookies

A cookie is a text file with small pieces of data and is used to identify your device as you use the network. HTTP cookies are a special kind of cookie that identifies specific users to improve and personalize the browsing experience. Cookies save information about a user’s session, the time you spend on a website, and use that information to remember specifics about the user the next time they visit the website. There are many different kinds of cookies with different purposes. In terms of the digital marketing industry, first-party cookies, third-party cookies, and session cookies are the most relevant.

Targeted Marketing

As you move from tab to tab in your browser, you are passing information back to other websites and the people who have access to those website’s data. Every time you visit a website, you should be under the impression that some of your data is being gathered by the provider. With this data, advertisers have engineered machine learning algorithms to determine a user’s interests, demographic, gender, relevant search history, and more information to determine the ad to show the end user. The process, Real-time bidding, is hosted by a published site and advertisers bid on a per-impression basis to determine which advertisement will be shown. The advertisements shown are different between users, and are largely dependent upon what information was parsed from their cookies. Real-time bidding has proven to be far more efficient than any traditional form of advertising, both for the advertiser and the consumer. In fact, consumers have come to expect brands to tailor their messages based on location, demographics, and interests. In a 2017 study, Accenture found that 40% of consumers had switched brands at some point due to poor personalization. Advertiser’s utilization of relevant data is critical to customer retainment as well as company growth. Optimization of these machine learning models will allow advertisers to better personalize their ads to the target audience, as well as save costs with more advanced bidding strategies.

Micro-Influencers

Since social media has gone mainstream, advertisers have been able to target ads based on what the user looks at across their accounts. The Tik Tok algorithm, for example, utilizes a wide variety of data points to get you the content that they believe is of interest to you. The more you use the app, the more data they get about you, and the more tailored to your preferences your experience will be. How advertisers benefit from this is they are able to find content creators whose viewers are in their target demographic. Larger brands who can afford the cost may still hire celebrities; however, smaller companies have realized that even though micro influencers may have a more “micro” outreach, their smaller size means that they are more interactive with their followers, meaning they have more influence. As your reach increases, it’s natural that your engagement falls, and vice-versa. To be specific, in a survey of more than 800,000 Instagram users, it was found that the ideal range for a micro-influencer is 10,000-100,000 followers. In this range it’s believed that the combination of reach and engagement is optimized. Active social media users increased by 10.1% over 2021, which equates to 424 million new social media users. Whether it be Snapchat, Instagram, Facebook, Tik Tok, or any other social media platform, the number of micro-influencers is going to increase dramatically with no sign of slowing down.

Chatbots

Chatbots are the internet’s version of customer service. Using Natural Language Processing technology, they are able to take a consumer’s input and understand, analyze, and compile human speech. Plenty of businesses use chatbots as a part of their marketing strategy to increase their trustworthiness. Also, as the demands for instant engagement and customer satisfaction continue to rise we are starting to see an increasing number of chatbots on websites. For example, as part of an expansion on their social media marketing, Match.com launched a Facebook Messenger chatbot called Lara as a “Match coach.” Since its inception, Lara has increased the number of registrations for Match.com by 30%.

Ethical Concerns

While advertisers collecting data can provide a host of benefits to consumers, there are drawbacks associated with this data collection. Consumers have begun to be more cautious about what data they choose to share with companies. A McKinsey study found that consumers are more comfortable sharing information with healthcare providers and financial services; yet, no industry has reached a data protection trust rating of 50 percent. Given the large-scale consumer data breaches that have occurred in recent history, this lack of trust is understandable. This has forced companies to become aware, and put an emphasis on how they are handling consumer’s sensitive data. Companies that are able to develop fast, safe, and easy to understand solutions to these problems have seen a competitive advantage over other firms in the industry.

Big Tech companies have been one of the largest culprits in terms of mishandling private user information. Senators are concerned that their unchecked access to personal information has resulted in the generation of in-depth profiles about Americans who use their products. Litigation surrounding Big Tech is almost non-existent, meaning that the Federal Trade Commission can’t go after them. The Sherman Act, an antitrust law passed in 1890, was not meant to be applied to our more recent technology and enforcement agencies are unsure how to move forward. Nothing that’s been done by these companies is illegal. When we sign up to use Facebook for example, we agree to the Terms of Service, Privacy Policy, and whatever else they put a checkbox for likely without even reading it. Consumers are beginning to realize that companies have no financial incentive to stop abusing the system. In April 2021, Apple released iOS 14.5 where users were able to choose whether they wanted to be tracked or not. A whopping 96% of users chose to not be tracked when given the option, showing that users value their privacy when given the choice. Even though users may choose privacy, the current system doesn’t operate under what the user’s choose, and instead consumers are forced to sign documents on account initialization to operate.

Big Tech Privacy Scandals

In recent years, data privacy and collection practices have been widely discussed and debated to prevent corporations from taking advantage of consumers. This is not the first time that advocates of privacy have spoken out; however, the amount of people discussing the issue has drastically increased. For example, in the 1990s Lotus MarketPlace had 30,000 out of 120 million of their consumers opt out of its database. When analyzed with the US population from 1990, those 30,000 people represent roughly 0.01% of the total population. As of 2020, roughly 69% of consumers are concerned with how their personal data is being collected and utilized by applications. This dramatic increase has mainly resulted from knowledge about how the data collection process works behind the scenes becoming more well-known.

Facebook is a common company for the public to bring up when talking about data privacy, and for good reason. They were, and still are, highly scrutinized for the handling of their data internally which has resulted in a plethora of legal issues. In March of 2018 it was found that Cambridge Analytica, a former British political consulting firm, had obtained private Facebook data from tens of millions of users. These documents contained proof that Stephen K. Bannon, former Trump Aide and Cambridge Analytica board member, illegally obtained all of this data. Mark Zuckerberg, Facebook founder, addressed this infront of the United States Senate, saying “what we know now is that Cambridge Analytica improperly accessed some information about millions of Facebook members by buying it from an app developer.” What followed was a stream of questions regarding Facebook’s business practices, political advertisements, user privacy, and much more. After the hearing the US Federal Trade Commission approved an estimated $5 billion settlement with Facebook following an investigation into their handling of user’s data. Steve Wozniak, an Apple co-founder, urged people to get off of Facebook over the dilemma. Public outcry for better protection of their data was heard long after the settlement and is still a prominent discussion to date.

That does not conclude Facebook’s scandalous user data activity. In 2021 an antitrust lawsuit between Facebook and Google was unsealed by a New York judge. The lawsuit exposed a secret deal between the two companies, whereby Facebook got unique identity data and a certain ad volume guarantee. In exchange, Facebook had to shift their ad network spend to Google instead of header bidding vendors. Internally the arrangement was called “Jedi Blue” and is an unequivocal example of market collusion and unfair dealing. This case is still developing, with the most recent development being Google filing a motion to dismiss the antitrust complaint. Adam Cohen, the director of economic policy at Google, stated in a blog post that “the complaint misrepresents our business, products and motives, and we are moving to dismiss it based on its failure to offer plausible antitrust claims.” There is much left to be done in this case, and not all the details are known; however, it will be interesting to see how this unravels and what the future holds for the two tech giants.

Government Involvement

Given the privacy issues that big tech poses, it shouldn’t be surprising that Governments around the globe are attempting to regulate what tech companies can and cannot do. Currently there are no regulations surrounding what these companies do with your data, whether they notify you of a data breach, or if an aftermarket seller is selling your data. Because these issues are so close to the consumer, 56% of Americans support more regulation for major tech companies. It was also found that 68% of Americans believe that large tech firms have too much control over the economy. A lot of regulation at this time is attempting to target antitrust practices and data collection rather than AI. The issue with regulating Artificial Intelligence is that a lot of times it will break down to regulating the data that the AI is trained on. Nobody truly knows how the algorithms work behind the scenes. What we do know is that training an algorithm on a biased/non-inclusive dataset leads to the AI having inherent biases. Amazon and Apple have each been caught utilizing a biased AI in the past few years. Apple’s credit card algorithm has been exposed for being discriminatory towards women. Men were more likely to receive a higher credit line and lower interest rates than women with the same credit score and credit history. Amazon on the other hand had an automated resume screening algorithm that filtered out female candidates. The important thing to note here is that neither of these companies were purposefully being discriminatory of women; however, the data sets that were training those algorithms clearly had a bias towards men, resulting in a biased algorithm. These situations are attempting to be handled by antidiscrimination legislation but there has not been much headway towards getting anything passed. Regulating how tech companies collect and use data, how they train their algorithms, and how they compete with each other are all going to be increasingly important as the world we live in gets increasingly virtual.