Really smart people are applying AI to just about everything, expecting better results. As we've reported intensively, marketing is no different. The industry is training AI to chomp through terabytes of data to seek insights that defy human analysis. So why is the focus group still around?
AI is a disruptive technology, changing the way marketers do business. It has moved from taking baby steps to taking full strides. It has not replaced the focus group, though it can complement one. There are still some things only humans can do.
Data is the Big Kahuna
Numbers can only tell you so much, but they still tell a lot. Data scientists can craft algorithms that can uncover “implicit information”, traits that customers may not tell you up front, noted Andrew Eichenbaum, Chief Data Scientist at Kahuna, a B2C digital marketing firm.
One example Eichenbaum gave was work done on an online recipe engine. The app proposed recipes to match user profiles. “There was a lot of complaints,” he said as “users did not believe what their profiles said about them.” One user said he hated sour food, yet the recipe he received for a barbecue fruit tart corresponded his preferences for “sweet” and “salty”, perhaps not understanding how these two tastes interacted.
This illustrates a strength of data: “You can look at dependencies,” Eichenbaum said, the interactions where one act leads to another. Can an action prompt a reaction? A user is shopping online, puts something into the cart, but does not proceed to checkout. Can a simple e-mail reminder prompt that shopper to complete the transaction? And if so, what is the right message? That can be tested.
Indeed, messages can be tested quickly in large groups, just by sending an e-mail. Thousands of responses can be reaped in just a couple of hours, Eichenbaum pointed out. “Outliers can be statistically significant”, for even a marginal variance in reply could represent a cohort large enough to be a test group, he said.
“Testing a wide range of messages is a good thing to have,” Eichenbaums said. Focus groups are “not dead yet,” he added. They are a good starting point because they can identify a basic idea. “If I have the ability to test in the real world, I prefer that myself.”
What AI can do to gauge customer preference can be pretty massive. Sentient Technologies relies on an AI foundation for its two main products: Sentient Aware, which staffs online stores with AI sales associates to help customers find what they are looking for, and Sentient Ascend, which automates web site testing with an eye towards increasing conversions.
While digital tools can paint the marketing landscape, the focus group can still offer insights within the boundaries of human comprehension. Sentient Aware can test web design for thousands of web pages, but “people may not get past the first few pages,” noted Jon Epstein, senior vice president for international at Sentient. A focus group can identify a problem with a web page, but testing the problem and its solution can be done digitally, he pointed out.
AI and focus groups each have their role. As for handling the sheer mass of numbers, “data will outweigh,” Epstein said. As for the gauging the subjective need for a product, “right now that is best coming from humans.”
“I prefer viewing AI as giving the user information to go alongside other information to make a better decision,” Epstein said. “Human intuition can be accelerated and advanced with these tools.”
How and Why
“For 20 years, the death of the focus group has been predicted,” said Brett Watkins, CEO of L&E Research, focus group specialists. So why is it not dead yet? Watkins described it this way: AI and “Big Data” are quantitative tools which can tell a marketer what has happened or is happening. But a focus group can tell you why consumers prefer a particular product or service, something AI cannot do.
“Say, 25 years ago, focus groups were the broad brush everyone was using to get at the ‘why' question.” Watkins said. Now it is just another tool in the tool box—still there, still useful. While focus groups are qualitative, AI is quantitative, he observed.
“The survey has been replaced —and augmented—by the Big Data revolution.” Watkins said. Purchasing behavior can be measured. The Internet of Things will generate even more data, especially when coupled with smart appliances, like a refrigerator that knows when you reach for the milk. All of that data can be analyzed. But that can't find out the whole story.
“Remember, we don't buy products in a vacuum,” Watkins said. A consumer will talk to friends and relatives when deciding what to buy, even visiting a “brick and mortar” store to see the product before ordering it online. That sale will be logged—and analyzed along thousands or millions like it. But the data cannot answer the question: “why do we behave the way?” Watkins said. You can't ask a person why they did something, which is why experienced psychologists are used as focus group moderators, “to explore the answer to get to ‘why'.” he added.