Is Synthetic Audience an Alternative to Customer Research? Experts Explain 

Customer research has got a new magic wand to play with, and it is AI. Yes, the synthetic audience — or you could simply call them AI agents of customer psyche. 

Imagine your brand plans to introduce a new product, and without reaching the actual target audience, you could secure reactions and feedback from the clones of these potential buyers. 

Would you consider it a bizarre phenomenon or plain AI slop? Reaching customers’ psyche behind purchasing and analyzing the data for better understanding should always be a priority for brands. However, it seems AI has penetrated this space too. 

It doesn’t matter whether this becomes a failed experiment or another AI bubble — brands and agencies are already introducing these “synthetic audiences” into their research systems. Let’s dive deeper into the term that’s been making headlines across major publications. 

Does a synthetic audience offer a fast, safe peek inside the mind of buyers?

Yes. That’s the whole purpose for now. A synthetic audience is simply a clone of your customers. They are AI-powered audience representations built from real data and deep insights, enabling a continuous dialogue with data — ready to answer even your trickiest questions. 

Nowhere should you confuse synthetic audiences with real buyers. These AI creations don’t use credit cards or BNPL schemes. Brands want these audiences because they give access to insights in areas where real humans may be inaccessible. 

Rob Chandler, Head of Marketing Science at Ogilvy, summed up their utility on a panel discussion organized by Ogilvy: 

“You can explore the boundaries of creativity by working with consumers during the creative process with help of these clones. Far further, they are synthetic — they don’t get offended, you cannot upset them.” 

“It is a safe space to ask them the unaskable and explore the boundaries of creativity.” 

Traditional research, surveys, and questionnaires take weeks or months to build out buyer personas. AI agents, on the other hand, are marketed faster and efficiently. 

But one data point from Ogilvy introduces a conflict: synthetic audiences can predict 85 percent of human behavior, leaving 15 percent of human unpredictability untouched. 

Brands can’t crack human unpredictability — ironically, that’s the part they should value most. Safe patterns are useless; the surprises are where real insight lives. 

How to create a synthetic audience — an experience

Jessica Cuebas, Director of Advertising Strategy at Workday, shared her experience creating synthetic audiences during the Ogilvy panel. Workday is an AI-powered enterprise platform for HR, Finance, Tech, and IT, primarily targets C-suite executives. 

She described these buyers as “hardly accessible assets.” These executives are busy, and brands cannot expect them to be available on call or text 24/7. She emphasized that this isn’t just a B2B issue — it spans industries. 

We all agree that quantitative research demands huge amounts of time and resources.

Cuebas explained that with the help of Ogilvy and WPP Open, her company created 20 synthetic audiences based on “the brains” of their customers — essentially, datasets layered to reflect actual buyer thinking. 

The scope of creating buyer personas with multiple variables is intuitive — and undeniably efficient. 

“Feeding the brain” approach

“Feeding the brain” simply means giving the synthetic audience the same diet of information a sharp strategist would rely on. Companies load it with first-party data, syndicated studies, secondary research, competitive scans, and even real interactions from sales calls and customer service logs. 

The model chews through all of this to simulate how people compare options, react to messaging, shift preferences, or develop new demands. It’s a compressed version of how human planners process the market. In short, “feeding the brain” is the act of loading every credible human and cultural signal, so the synthetic audience behaves with grounded, real-world logic rather than fantasy. 

A word of caution to brands

“We do not want to trap ourselves into the synthetic stereotype spirals.” 

Dan Bennett, Global Lead of Behavioral Science at Ogilvy Consulting, offered a necessary caution for brands eager to LARP these synthetic audiences into their systems. 

He’s right. Overreliance is a continuous mistake brands keep making. They cannot assume customer insights are now instantly available at their fingertips. 

Synthetic audience requires constant feeding of new data, behavior changes, and evolving business landscapes to stay relevant. Without ongoing updates, they become stale fast. 

Another challenge is AI’s tendency to generate favorable or biased answers. Rob Chandler addressed this head-on: 

“We are not solving all the world’s research problems. Humans have biases. Every study has some essence of bias in it. We make billion-dollar decisions in a room full of 11 people. We are not solving the challenges of biases or favourable answers. We are going to look at them with different lenses.” 

Chandler outlined how data should be layered: start with demographics, layer psychographics to mimic how the real audience thinks and behaves, and finally add contextual understanding. 

That last layer, context, is what gives the synthetic audience real power. 

And in this age of AI, we still cannot undermine the value of real humans. Synthetic audiences are not here to replace human research. 

Cut to the chase

Synthetic audiences dominated this year’s final quarter headlines for agencies and brands. These AI agents promise strong potential for the coming years — but the challenges are running right alongside the hype.