One of the most potent sources of information for advertisers and marketers is how consumers discuss a brand when they’re being most honest. Sometimes, this happens in online product reviews, although research shows that those tend to over-represent extreme views. It also sometimes happens through social media posts or videos in which people mention brands. So, since the advent of social media, brands have turned to “social listening,” in which marketers track mentions of a brand across various social platforms, and look for common keywords, themes, or ideas.
While this has value, social listening also has major limitations. For one, it doesn’t allow a brand to ask questions. But perhaps even more significantly, it doesn’t yield representative findings. Most Americans are on social media, but many are “lurkers,” not saying anything. “A minority of extremely active tweeters produced the overwhelming majority of all tweets made by U.S. adults,” Pew Research found. And even among those who are active, how many actually discuss brands? When’s the last time you posted about your laundry detergent?
In our work at Glimpse helping organizations discover what different groups of people think and feel about various brands and ideas, we’ve been using AI, including new generative AI tools. We’ve found that this emerging technology can transform the process of collecting feedback from consumers, reaching more people across more demographics than ever before, and learning more about what they really think. Here’s what leaders need to know.
What Leaders Need to Know
Generative AI tools can rescue findings from the “trash.”
Most organizations have reams of survey responses buried in their files, and more likely than not they contain untapped value. As one study put it, companies notoriously “rest on huge amounts of unused data treasure.” This is especially true of qualitative data — that is, consumers’ responses to open-ended questions. The chief insights officer for a Fortune 50 corporation told us his company has loads of qualitative data “sitting in a trash can.”
Why does this happen? Because reading through all that text has been a long, difficult slog. And even when marketers did read some of these responses, there wasn’t much they could do with them. So, instead, they’ve often focused more on quantitative responses, in which consumers rate their feelings on a numerical scale or select from multiple choice answers.
Generative AI tools have enhanced levels of natural language processing that go beyond what was previously available. These tools understand the conversational language people use, and businesses can “teach” them to understand slang terms.
So generative AI can scan through huge amounts of text and immediately discover recurring sentiments, emotions, perceptions and experiences that consumers describe. Even better, these tools can break down qualitative responses into demographic categories, looking at the kinds of things people in various age, racial, ethnic, or other groups more frequently say about a brand.
And given what these tools find, they can point to communities that have been underrepresented in survey responses.
Generative AI tools also allow you to dig deeper for insights.
Generative AI tools can also analyze the text inside surveys and discover which kinds of questions are most likely to evoke responses. This technology can then help develop questions worded in ways likely to maximize response rates through natural, conversational style. And when those responses come in, the tools can instantly piece through the new responses and highlight new, more advanced findings.
Take, for example, our work with Ayzenberg, an ad agency focusing on the video gaming industry. Generative AI analyses of previous research showed that members of various minority groups faced numerous negative experiences in these communities. Ayzenberg wanted to learn more. So, we developed a series of survey questions and sent them out to women, BIPOC, and LGBTQ+ players.
Responses came in quickly. Generative AI tools we use read through all of the text in the open-ended questions and provided crucial insights within minutes. Among the findings: 70% of respondents to this survey said they had witnessed toxic behavior among players within the context of games, such as during multiplayer games or in forums in which gamers chat.
These tools instantly gave us breakdowns by audience segments. We found that BIPOC respondents often experience being called racial slurs and derogatory names. Members of the LGBTQIA+ community frequently described being targeted for their gender or sexual orientation, and being called homophobic slurs. Many women reported being harassed, insulted, or fetishized. The generative AI tools were able to show us this despite the fact that respondents used different terminology to describe the same types of situations. Some video game companies are now using these findings to develop strategies for counteracting these problems.
What Companies Can Do
Here are three simple ways organizations can use generative AI tools to turn qualitative data into powerful market insights:
Tailor your generative AI tools to meet your organization’s specific needs.
Teach your generative AI tools what you’re looking for. Give them context and guidance. Platforms can provide sophisticated, and automated ways to tailor generative AI outputs in this way. But you can start by simply describing in natural language what actionable insights you need, and the tools will seek those out.
Feed them your existing data.
Take all those surveys you may have sitting in old files, filled with customer responses that no one knew what to do with. Include any high-quality, representative, and salient data that can add value. Let your generative AI tools read through all of it and provide you with a brief summary of recurring insights and problems that your target consumers express.
Have them make suggestions for your next survey.
Generative AI tools can point out demographic groups that may have been underrepresented in your survey findings. They can even suggest topics to look into or hypotheses to test.
Once you send out the new survey and the results start to come in, repeat the process. Because generative AI is designed to keep learning at all times, its power snowballs.
These tools can also track how expressions of emotion — from suspicion to excitement — change over time, and whether they’re becoming more positive or negative. This helps brand leaders determine whether their efforts are working.
To be clear, the power of generative AI does not negate the value of social listening. A holistic solution combines approaches, including third-party social data gathered by listening platforms and an agile system to collect first-party data.
Ultimately, the goal is to get the clearest possible picture of the real world. The more organizations use generative AI to understand the consumers they’re trying to reach, the more successful they’ll be in tapping into its game changing power.
