Business frontline evolution driven by generative AI: utilizing AI to analyze user needs and values | Insights

Business use of generative AI is progressing rapidly with the emergence of services such as ChatGPT and Microsoft Copilot. Much of the application of AI, however, has been limited to streamlining the work of individual people by, for example, summarizing or organizing text and data, meaning that the scope of use of generative AI remains limited.
At ABeam Consulting, we are expanding the scope of the various ways to utilize generative AI as a means of solving business problems, including but not limited to the streamlining of individual people’s work, and developing new ways of using the technology. One example of such utilization is analyzing user needs and values.
In this insight, we go over the key points and major concerns of using generative AI, based on a case study into the utilization of the technology in analyzing target user needs and values as part of a project to formulate and implement a brand strategy that we supported.

Utilizing Generative AI to Analyze User Needs and Values

Understanding the needs and values of target users is something that is required in many projects at the strategy formulation stage.
This goes for both brand strategy formulation and marketing measure proposals. Due to the fast-changing nature of recent markets, it is considered desirable for companies to very speedily and iteratively perform surveys of target user needs and values, and incorporate the findings into their strategies and measures. Companies normally perform surveys by recruiting people who fit with their target audience from among the general public and then interviewing them. However, the issue with this approach is that it imposes costs in terms of time and money on companies while they recruit participants and wait for survey results. Even if companies hand the data gathered by those surveys off to generative AI for analysis, it will still generally take at least two months to do the whole process. There are also increasing demands being put on companies to take into account such as personal information management and ethics investigations. So, because regularly surveying real people and correcting strategy and marketing measures based on the findings incurs considerable time and money, companies will find that the barriers to regularly continuing the survey are quite high, even if they followed it initially.

Amidst this context, ABeam Consulting has opted for a new method in which, instead of real people, it prepares a generative AI with the attributes of the target users, and gets responses to questions posed to the generative AI. Specifically, the program is designed so that the generative AI is preset with target user segment information defining factors such as age, gender and income, with it being set to respond to all questions as though it were a human who had said attributes. As the generative AI will respond with representative/average opinions for a person with the given attributes, companies can receive the same results as if they had interviewed the majority of people with those attributes.

Companies are thus able, to a degree, to perform surveys of the needs and values of their target users in a way that is interactive, while cutting down on some of the costs in time and money associated with such a survey (figure 1). Making use of this approach returns responses that pinpoint what surveyors want to know. This allows them to finely test hypotheses around strategy and marketing measures at the drafting stage by deepening their understanding of users through repeated, detailed questioning.

Figure 1. Comparison with regular survey methods

Case Study of Generative AI in a Project to Formulate and Implement a Brand Strategy

Here we present how generative AI has been utilized to analyzing target user needs and values in a project to formulate and implement a brand strategy and marketing measures.

To gain recognition and support for the brand among the users being targeted, the company needed to discover the messages and values that would resonate with those demographics, and to serve those messages via channels that would be able to capture them. It was thus critical for the company to properly understand the needs and values of its target users. But, as stated above, the regular survey method of interviewing real people would incur a certain of level of time and money. In this project, the client was seeking to move extremely quickly from brand strategy formulation to executing marketing measures, so they were focused on getting results through a quick and agile cycle. We thus opted to use generative AI.

In this project, we first performed a macro-analysis, then sorted and defined target users into four segments according to age, income and where they lived. We performed desktop research about each segment’s needs and values, then formulated hypotheses. Having done so, we used generative AI to do the following.

1. Testing hypotheses about the needs and values of each segment

Firstly, we set up the generative AI by creating prompts (commands to the AI) to explain the attributes of the four segments. The prompts were made up of the ages, incomes and where they lived that defined each segment, with the generative AI being set to respond to all questions as though it were a person who had said attributes.

Next, we drew up questions to ask the generative AI to test our hypotheses about the needs and values of each segment. Because we found that, due to the nature of generative AI, it tended to err on the side of giving affirmative answers to closed yes/no questions about hypotheses, we paid special care to wording of our questions. For example, rather than asking questions that immediately end with something like “are you happy with your current lifestyle,” we presented the program with antagonistic options at the same time by asking questions such as “are you happy with your current lifestyle, or do you feel there are some problems?”

Finally, we asked each of the questions we prepared to the generative AI set according to the attributes of each questions, then received responses (figure 2). Where necessary, we added questions and had a back and forth conversation with the generative AI, analyzing whether our hypotheses were correct, and, if they were wrong, how we could go about improving them.

Figure 2. Visualization of the process of testing hypotheses about needs and values and how to use it

2. Exploring the boundaries of the needs and values that appear among segments

To consider specific marketing measures that fit with the characteristics of each segment, following testing of the hypotheses about each segment’s needs and values in 1, companies need to further raise the resolution to figure out under what conditions the needs and values of each segment vary. To understand where the boundaries appear in the differences among the needs and differences of each segment, we once again used generative AI.

Firstly, we drafted multiple choice questions that got at the needs and values of users from a number of angles. For example, we asked questions such as, “when you purchase products related to ○○, do you make decisions with a focus on speed, or do you carefully compare a variety of options?”

Next, as in 1, we set up a generative AI with the attributes of each segment, asked the multiple choice questions we drafted for each segment, then compared the responses. When the responses differed across all four segments, we determined that those responses were a characteristic that expresses a difference in needs and values across the segments, and we recorded this fact. When we found that the responses were the same across all four segments, we determined that this does not stand out as a characteristic of the segments. We then adjusted the details of the question openings and options until a difference in responses appeared.
By thus repeatedly comparing questions and responses, we clarified the key characteristics of each segment (figure 3).

Figure 3. Process for searching for and exploring the differences in needs and values of each segment

Utilizing Generative AI to Solve Problems in Business and Society

In the project presented here, we managed to shorten a process that normally takes around three months to around two weeks by utilizing generative AI to perform surveys on the needs and values of target users. We were also able to ask additional questions when further points we wanted to ask about arose in the process of the study. This project is set to enter the phase where the company implements and evaluates specific marketing measures. we plan to likewise utilize generative AI to study how to improve those measures, having evaluated the reactions of real target users and updated its segment information based on those results. We are also working on utilizing the technology in the strategy formulation phases of other projects.

With that said, the responses of generative AI remain strictly representative responses for the segments in question. Companies should be aware that this method, therefore, does not act as a complete substitute for traditional interviews and surveys targeting real participants. In responding to questions, what generative AI does is gather relevant information and produce plausible sentences. Thus, it is reasonable to think that it can check the opinions of the majority on a hypothesis, when used to perform surveys on the needs and values of users. Conversely, however, it also cannot get minority or stand-out opinions. There are cases when, in planning new services, it is more necessary to get hints from extreme, minority opinions, rather than checking what the majority opinion is. In such use cases, it is more appropriate to employ the traditional method of conducting in-depth interviews with real extreme users, rather than relying on generative AI (figure 4).

Figure 4. Scope of what generative AI can be used as a substitute for

Generative AI is still seeing the scope of its utilization expand in a variety of directions, and could prove useful to solving problems in business and society. However, as described above, it cannot do everything, so if companies do not use the technology correctly based on its characteristics and limitations, they will not get the desired impacts.
ABeam Consulting has a track record of using generative AI in ways that contribute to solving various social and business challenges. The scope of potential utilization of generative AI is broad and varied, including streamlining of internal inquiries, ameliorating employee sleep debt, transferring skills from veteran employees and enhancing credit checks . We are also well-versed in the characteristics and limitations of generative AI, enabling us to propose suitable use cases in response to client challenges, so please feel free to get in touch with us to discuss your situation.
ABeam Consulting continues to utilize technology in our efforts to solve social and business challenges.

Insights TOP