Synthetic Research Isn’t the Future—It’s the Shortcut

In a world where insights are currency, marketing leaders are being asked to do more with less. Less time. Less budget. Less margin for error.

The truth? You’re increasingly under pressure to get results from consumer research quickly. You might not have the budget to re-interview 500 people. And if you’re being honest, you’re still not sure how Gen Z feels about your new brand platform.

Enter synthetic research. Not as a replacement for tried and true research methods, but as a tool that helps marketing leaders move faster, with confidence.

And that matters, especially now.

Synthetic research is the practice of using AI to simulate feedback using what we already know about consumer behavior, often layered with existing survey results, sales data, and media trends to improve the quality of the model. Once the model is hydrated (taught) to act as our audience, we can use it to test how target audiences might respond to real-life scenarios, such as testing creative direction, validating messaging, and exploring white space. And here’s the added benefit: We can do this to get timely, directional insights to help evaluate next steps — whether that is investing in full-scale traditional research or going back to tweak concepts before retesting.

Below are key differences between traditional and synthetic consumer research.

Use it for direction, not for depth—to better understand category perceptions and receptivity to new brand messages, not brand perceptions.

Synthetic research won’t replace live qualitative or large-scale quantitative studies, because the existing data sources it draws from often don’t capture meaningful brand perception for most brands. Without enough high-quality, brand-specific data in the public domain, synthetic methods can’t provide the same depth of insight as direct audience research — but it’s an ideal option when speed, feasibility, or iteration is key.

Use it to:

• Fill sample gaps with hard-to-reach audiences
• Pressure test early-stage messaging
• De-risk new campaigns with directional input
• Explore sensitive or privacy-sensitive topics

Don’t use it for:

• Brand awareness or perception audits
• Emotionally layered, high-context insights
• Regulatory or market-sizing studies

Before you add synthetic research to your insight stack, ask:

• What decisions will this inform?
• Is this a directional need or a precision need?
• How will outputs be validated?
• How are data privacy and ethics handled?
• How can we use primary research, first-party data or publicly available data to train the model for multiple use-cases?

These questions separate shiny tech from strategic value.

Not all synthetic tools are created equal. Ask your partner:

• Where does the training data come from?
• Can we securely add de-identified first-party data and primary research, with assurance it will only be used for our study?
• How do you avoid bias amplification?
• Can I validate your panel design or prompts?
• What’s your plan to maintain model accuracy over time?

If they can’t answer clearly, walk away.

At Lewis, we’re actively reviewing partners to explore message testing with synthetic panels and sample boosting for small or segmented quant studies. We are also strong advocates of using synthetic feedback to sharpen creative or campaign strategy. To put it simply, we’re not replacing our toolkit—we’re expanding it.

Synthetic research isn’t perfect. It’s a tool that is actively evolving and improving, but great strategy still depends on human instinct. On asking the right questions. On seeing the nuance that synthetic models can’t.

But it’s perfectly suited to meet the needs of marketing teams — fast, interactive, and insight-led.