August 5, 2024
Can Artificial Intelligence Improve Marketing Decisions?
Philip Kotler
Will Artificial Intelligence (AI) prove to be a blessing to Marketing?
I am hearing many conflicting opinions about AI.
There are negative opinions. The neuroscientist Erik Hoel has called it “A.I pollution”. Others call AI “something like noise.” Sequoia Capital says that current AI investments have been running far short of expected returns. Goldman Sachs has echoed the same concerns.
Other skeptics have pointed out some occasional errors in AI output, such as calling Barak Obama a Muslim or that 17th President Andrew Johnson earned 13 college degrees between 1947 and 2012. Hopefully, these are rare exceptions to the generally correct AI outputs.
Those who are positive about AI’s impact on business cite how much more productive AI solutions are compared to “human-only” solutions. AI can speed up the search for solutions to various human diseases. AI can beat the best humans at a game of chess or Go. AI can turn out a preliminary 30 second video story ad in five minutes that might have taken an ad agency a week and many dollars to prepare.
A 2018 McKinsey analysis of more than 400 advanced use cases showed that marketing was the domain where AI would contribute the greatest value. In August 2019, the American Marketing Association reported that AI implementation had jumped 27% in the previous year and a half. In 2020 Deloitte reported that AI could help in the following three marketing areas: enhancing existing products and services, creating new products and services, and enhancing relationships with customers.
To move our understanding of AI’s potential for marketing, let’s distinguish between General AI and Generative AI (such as Chat GPT).
General AI
Companies have told their marketing staffs to learn AI and to propose applications that would speed up and/or improve the quality of its decision making. The company needs to identify some important decisions in that business and determine whether a better decision could be made by a machine solution or at least by a person+machine solution.
Consider a bank that makes lending decisions. Normally, one of the bank’s executives specializes in approving lending decisions. This executive has spent years making lending decisions. He has developed a mental framework for making his lending decisions. Most of his past decisions were good but he also made a few bad decisions where the bank lost money. Can the bank build an algorithm that uses this executive’s mental framework, but improves it in a way that had the algorithm been used in the past case, the company would have achieved even better results. If this new algorithm is then applied to new lending cases, and the company achieved even better results, tell the bank manager that the bank will use the algorithm and the bank manager should move to making more strategic decisions.
Generative AI
On November 30, 2022, A company called Open AI announced to the public that it had created ChatGPT. Chat GPT is based on building a large language model (LLM) capturing a great deal of the world’s past conversations. A copy writer can type in a prompt, such as “write a short essay on the planet Mars.” In a few minutes, the printer would print some pages presenting a clear description of Mars and its features. Or the prompt may be: “write a love poem about Mars.” Or “draw a picture showing Mars laughing at the other planets in the earth’s galaxy.” Or “create an ad for the sale of 3 pound rocks found on Mars.”
The prompt can include the desired length, format, style, level of detail and language. After viewing the output, the marketer may want to submit a more refined prompt to improve the desired output.
Stages in Installing AI in the Company
A company might be enthusiastic about installing AI but needs to know that it will take a long time to become proficient in AI. Why? The company’s marketers probably lack training or experience with AI and must go through a cycle of learning. The company needs to hire an AI expert into the department or hire experts on an advisory basis. The marketing department meanwhile must have a good working relationship with the Intelligence Technology (IT) group that will manage and collect the needed data.
One early step would be to prepare a chatbot that can answer a lot of frequently asked consumer questions. The chatbot would operate much like Siri or Alexis in answering questions posed by consumers.
The next step is to identify some routine marketing procedures that can be done faster and better by automation. Any filing work in the department can be done more efficiently with AI with less chance of human error.
Another step is find an important area of repetitive decision making and create an algorithm that can make better decisions than humans can make. Over time the algorithm will be self-learning and improving.
Another early step is to automate high-speed decision making in programmatic ad buying.
Still another use of AI is to make individual buying recommendations based on each customer’s information. This is called setting up a recommendation engine.
At some point, the company will move from “task automation” to “machine learning (ML).” ML is a subset of AI that allows computers to analyze and interpret data without being explicitly programmed. ML assists humans in solving problems efficiently. The algorithm learns and improves performance and accuracy as more data are fed into the algorithm.
The company can’t do ML until it collects a huge amount of quality data on customer buying behavior. The machine will search the data for patterns of behavior and their influencing factors in the hope to gain valuable consumer insights.
What the firm wants is a sales process for managing individual accounts using real-time geolocation data and detailed information on the customer’s reading habits, shopping habits, and decision making routines. The marketer needs to know what brands the customer buys, what sizes, with what frequency. The sales system will suggest sending reminders or special offers at the right time and place in a highly personalized way. The system will also suggest when a customer would be ready to move to a higher price item (upselling) or should add another buying category (cross-selling).
AI can be used to write marketing content that can be disseminated in social media posts, email copy, blog posts, and captions. In addition to written content, AI can be used to generate images, audio, and video.
AI can be used to search for new customers. The characteristics of the company’s current customers (gender, age, income, etc.) can be used to search for potential clients with similar characteristics. AI can be used to sight the most persuadable clients and to choose the best messages to send to them via chat, phone, email, or video.
Conclusion
Artificial Intelligence (AI) has vast potential in marketing. It leads to collecting valuable data and extracting new insights on buyer behavior. It enables the development of software for data and sales management. It leads to the development of algorithms for better decision making than would have been provided by only relying on human judgment. It leads to the use of GPT to quickly develop messages, images, and virtual ads.
For firms without AI, the time is now to begin the AI journey. AI will ultimately affect the management of all elements of the marketing mix, namely product, price, place, promotion, branding, and marketing research. Some marketers will have to move to new skills and jobs or be replaced. There will be high costs, mistakes and some chaos but it will be worthwhile in terms of the benefits and the increased chance of surviving in this time of radical innovation and rapid change.
For a helpful book with more detail on AI and machine learning (ML) as well as other new developments useful in marketing such as robots, drones, blockchain, the internet of things, and the metaverse, see V. Kumar and Philip Kotler, Transformative Marketing: Combining New Age Technologies and Human Insights (Palgrave Executive Essentials), 2024.
