Virtual Personas by Savanta

Success with virtual personas relies on more than the artificial. That’s why Virtual Personas by Savanta always centre real-world customers, experts, and context – even down to human drivers like emotions.

But when the personas are virtual, the confusion is real.

We consistently here from our clients: “Can virtual personas truly support speed and growth? Or are they low credibility hype, the latest version of ‘garbage in, garbage out’?”

It’s a fair question. But here’s what’s clear: virtual personas can enable rapid, lower-risk growth. That is, if they’re implemented to solve problems without causing bigger ones. Simple goals sometimes require complex solutions.

Where heritage meets innovation

For the past 20+ years, Savanta has been supporting critical market research for brands across sectors and the globe. Our legacy is grounded in qual and quant, rigor and methodology. Of course, we’re also committed to innovating, so our customers can always best understand their customers.

Recently, our researchers, technologists, and data scientists have been working behind the scenes on the latest method for quickly gathering actionable insights: enter virtual personas.

And guess what? We started with the weaknesses of AI-enabled personas, so we could engineer a better solution for you from the ground up.

Conversational AI does not mean credible AI

Successful adoption of virtual personas is about more than just new AI. Truly understanding customers at speed starts with a foundation of rigorous research. ChatGPT or other large language models (LLMs) can generate text that sounds like a person. Conversational, yes. However, not informed enough to be useful, let alone credible.

Without quality data to train your AI tools, virtual profiles are just “skin deep” at best. Savanta’s internal experts and customers regularly mention the following challenges when using unreliable virtual personas;

• “Garbage in, garbage out”: Programmers have this saying about the consequences of bad data. It’s especially relevant for virtual personas, which depend on quality inputs for training and validation.
• Need for transparency: Technologists warn against “black box AI” where you can’t see inside to understand where an output comes from. These false outputs are called hallucinations, something you definitely don’t want informing your market choices.
• Lacking human nuance: Studies show that LLMs have a tendency to gravitate towards the middle when representing an audience. This sycophantic clustering is driven by telling you what you want to hear, not what you need to know based on actual human emotions.

The business need for human-focused AI tools

To accurately synthesise customer sentiments and truths, your brand needs to keep people at the core of your research. How? By proactively maintaining human focus even when using automation. For AI to be able to augment research by providing faster and richer insights, it must meet certain needs for quality and credibility.

So, how do you keep humans at the centre of your research?

1. Use real data, reflecting your real customers. Tangible insights can only be derived from data gathered in the real world. Quantitative survey responses, qualitative focus group interviews, and other data need to be the core of your virtual personas.
2. Layer in real-world sector expertise. In addition to brand reputation, you also need to train your virtual personas on what your sector is experiencing. If your organisation could use help here, prioritise finding a research partner with in-house expertise. This pays off both when training personas, as well as interpreting outputs and insights.
3. Keep an eye on the macro context. Reflecting – let alone eventually predicting – the human perspective is also reliant on macro context that impacts human choices. For instance, how is economic uncertainty impacting your banking customers’ risk tolerance?
4. Focus on the human factor of your virtual personas. ChatGPT and other LLMs can generate convincing-sounding text when prompted. But alone they can not reflect human emotion and nuance the way your customers do.

How Savanta’s legacy enables your own, too

Today’s marketers and insights teams have to do more with less, including less runway to gain insights. To help our customers, Savanta has designed virtual personas to accelerate research and augment capabilities.

To bring your AI-enabled personas to life, we start with our 20+ years of traditional market research, from weekly BrandVue surveys to in-depth fieldwork. Each persona is a synthesis of trustyworthy data gathered from your customers. Of course, they’re also levelled up with broader context like public or wealth sector experience, as well as the wider political and economic landscape.

What brings all this AI-accelerated output to life?

In short; Savanta’s proprietary understanding of the human drivers behind choices.

When research showed that other virtual personas struggle with emotional questions and responses, we knew that was the missing link. By combining in-depth emotional and value-based data, Virtual Personas by Savanta are anything but artificial.

Curious to get ahead of the curve and apply virtual personas to your research? Join our mailing list to learn more about Virtual Personas by Savanta and ensure you don’t miss the official launch date.