Generative AI reached a peak in the hype cycle this year, with dramatic predictions that ranged from handwringing about the doomsday of the knowledge worker to viewing GenAI as a worldwide panacea lovingly dropped on our doorstep. We’ve seen this same dialog play out specifically in the insights industry with almost equal hyperbole, particularly when it comes to the use of synthetic respondents.
The rise of Gen AI has led to the rapid development of synthetic respondents that can generate human-like responses to questions based on available datasets. In an ideal world, these models can replace human participants on certain questions, shorten interview length, and speed up the discovery of insights across the research spectrum.
While synthetic respondents can augment available data and accelerate decision-making, the human voice will always be irreplaceable for certain business decisions. Now the challenge for marketing leaders is to find the right balance between the efficiency of synthetic respondents and the essential insights provided by real humans. Below, I offer a framework for how to think through that decision.
FOUR QUESTIONS FOR DETERMINING WHEN TO USE SYNTHETIC RESPONDENTS
Synthetic respondents unlock rapid data collection, cost-efficiency, and scalability that help accelerate preliminary research and inform business decisions. However, they should be used carefully to ensure that they complement, rather than replace, the depth and nuance of human participants.
Here are four questions to ask when considering whether to use a synthetic or human respondent set:
1. IS THE QUESTION BUSINESS-CRITICAL?
When evaluating whether a synthetic audience is right for you, consider the criticality of the question you are trying to answer. For example, if responses will be used to define your organization’s strategy for the next five years, think twice before relying on synthetic responses that offer minimal sourcing of logic or reason or the ability to follow up on specific questions with clarity.
For business-critical questions, it is worth the time and cost to leverage real human feedback in a variety of forms (qualitative and quantitative research, behavioral data, financial data, etc.). This ensures strategic decisions are rooted in authentic, real-world insights, guaranteeing the highest level of relevance and reliability for your organization.
2. HOW SPECIFIC IS THE AUDIENCE?
If you’re a soft drink distributor looking to gauge broad consumer trends and tastes, synthetic responses can function as a time- and cost-effective option during the initial phases of analysis.
But if you’re a pharmaceutical manufacturer looking for more specific insights on how to approach a certain care protocol, you need responses from an experienced professional. A human voice is indispensable when specific audience insight is needed, ensuring precise and resonant insights that synthesized responses trained on generic data sets cannot provide.
3. WHAT IS THE STAGE OF THE RESEARCH JOURNEY?
Multi-phase journeys that begin with a long list of opportunities and narrow to a more focused shortlist are good candidates for synthetic responses.
In the past, marketers turned to humans for input during the initial stages of this type of process. However, with synthetic responses, you can quickly and cost-effectively capture insights as a preliminary step before you dive into more in-depth qualitative engagement on user experience, user testing on products, or advertising testing.
The key is to strategically identify where AI can seamlessly integrate into an iterative and comprehensive marketing strategy to best capture the consumer’s voice.
4. HOW COMPLEX IS THE QUESTION?
Binary or multiple-choice questions that you might ask a general population audience are easier for a synthetic respondent to answer than open-ended questions. GenAI can provide answers to open-ended questions, but if the question you’re asking will impact the future of your business, you’ll need to interact with humans to receive a more discursive answer.
For instance, questions like “Which U.S. sedan are you more likely to buy?” are ideal for capturing a synthetic response that sits on top of a relatively straightforward dataset. But to understand how a human will interact with a relatively complex piece of technology, you need a real person interacting in a qualitative atmosphere.
THE HUMAN VOICE REMAINS CRITICAL
AI advancements, like synthetic respondents for market research, have created opportunities to augment and accelerate the human voice. Both at the market research stage and throughout the marketing journey, marketers benefit by receiving feedback more quickly and affordably than if they were only working with human respondents.
However, AI can never fully replace the human perspective. For nuanced and business-critical insights, you need input from actual users who bring a depth of experience, emotions, and contextual understanding that AI lacks.
As CEO of Sago, Reed leads global strategy and business growth and is focused on digital transformation and customer-centric solutions.