Active-duty service members often have limited free time, making it difficult for researchers to collect reliable survey responses. This creates a major challenge for organizations trying to understand their needs and preferences.
For Kathleen Myers, assistant vice president of research strategy at Navy Federal Credit Union, gathering feedback is essential. The institution relies on member insights to improve services for military personnel, veterans, and their families.
Synthetic data, which uses AI-generated responses to mimic real-world behavior, is emerging as a promising alternative. It offers researchers a faster and potentially more scalable way to gather insights.
Interest in synthetic data continues to grow. Research from Qualtrics found that 41% of market researchers already use synthetic data, while 62% plan to adopt it in the future.
Navy Federal Explores Synthetic Data Potential
Myers believes synthetic data could help solve the challenge of reaching active-duty service members. Once models are properly trained, they can generate feedback quickly and at a larger scale.
To test this potential, Navy Federal partnered with Qualtrics on a pilot project. The goal was to evaluate how synthetic data compares to traditional third-party survey panels.
According to Myers, the pilot focused more on learning and experimentation than immediate business outcomes. The project aimed to better understand the strengths and limitations of synthetic responses.
Synthetic data offers clear advantages, including lower costs, faster results, and reduced bias. However, experts still debate whether the technology is ready for broader decision-making use.
Synthetic Data Performs Well in Functional Responses
Myers noted that synthetic data performs best when answering practical and rational questions. It tends to provide strong responses for functional topics and decision-making scenarios.
emotional questions remain more challenging. Synthetic responses can struggle to fully capture human feelings, emotions, and deeper motivations.
Navy Federal and Qualtrics tested synthetic data on consumer attitudes toward trust and financial services for a potential credit card package. The goal was to measure how closely synthetic responses matched human responses.
Qualtrics surveyed 501 traditional panelists and generated 498 synthetic responses. Across most questions, both groups produced highly similar results, with mean scores differing by only 0.25%.
Read :Â How USPS Shippers Can Handle Pricing Changes in 2026
Key Differences Revealed in Human vs Synthetic Responses
Despite strong similarities, some differences emerged between human and synthetic responses. Synthetic participants were more likely to prioritize efficiency and financial benefits.
At the same time, synthetic data underestimated the importance of empathy in decision-making. This highlights one of the key limitations of AI-generated responses.
Qualtrics also found that synthetic data may reduce common human biases. For example, human respondents often show acquiescence bias by agreeing with statements more frequently.
Social desirability bias was another major factor. In one survey question about reluctance to apply for a credit card due to fear of rejection, 41% of humans agreed compared to only 21% of synthetic responses, showing a significant 20-point difference.
