Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

Your consumer persona is a simulation

Marketing teams need to move past knowing who their customers are to understanding how they'll behave.

[Photo: Getty Images]

Profile photo of Chad Reynolds

Every day, we make business decisions we believe are in the best interests of the customers we serve. We imagine what they might think, do, and how they might react. Some of us create visual artifacts that bring customers to life as personas. While these are well-intentioned, they are influenced by the past rather than the future of how consumers might behave.

PERSONAS AS THE STARTING POINT

For decades, the customer persona was the answer. Personas worked, and they gave teams a shared language, a way to make decisions that wasn’t purely gut instinct. But they encouraged the oversimplification that a static description was the same as real understanding.

A persona can tell you what someone values in the abstract. It cannot tell you what that same person will do when they’re tired, skeptical, hurried, worried about money, influenced by a peer, or forced to choose between two values they genuinely hold. Our recent white paper explains that these traditional personas are static profiles and cannot account for the collision of these factors. But those static profiles became the unquestioned standard the marketing industry has been relying on for decades.

A static profile describes who someone is, but it cannot compute what they will do when those layers press against each other, which is exactly when decisions get made. There is a meme that illustrates this. It shows a description of two men, born in 1948 in the United Kingdom. Both live in a castle, have been married twice, are wealthy, and famous. Surely, they occupy the same marketing segment and will reliably respond to products and services in nearly identical ways. Except that the two men are King Charles and Ozzy Osbourne.

Personas break down because they can’t see what lies beneath their assumptions.

LANGUAGE MODELS AS A MOUTHPIECE

You have likely experimented with using LLMs to simulate personas, whether prompting the model to act as a trend analyst or as a consumer segment evaluating potential product launches.

Initially, the feedback generated seems convincing. Pressing further with follow-up questions usually reveals a common limitation: the LLM rapidly sheds its adopted persona and reverts to generic, average responses—the statistical mean rather than the sharp, distinctive insights required for strategic planning.

Despite widespread expectations that training massive language models on vast human text will enable them to accurately simulate consumers, this approach ultimately fails. While these models excel at mirroring how individuals articulate preferences online, they completely lack an internal psychology. The authentic consumer experience—marked by fatigue, skepticism, financial constraints, and complex decision-making—is rarely captured in public internet text.

Language prediction and human simulation are different jobs. That’s why it’s important to marry AI-driven processes with real human interactions to achieve lasting, durable results.

To close this gap, we developed something called People Models, which are built on an advanced Large Persona Model (LPM) that is customizable, multifaceted, and reflective. Its training data is drawn from real conversations where people reveal how they arrive at decisions, including all the friction, contradictions, and resistance that never makes it into surveys.

The model computes behavior from the simultaneous intersection of three things: who the person is psychologically, where they live in space and time, including what limitations they have, and how they make decisions, especially when those decisions require complex dependencies.

When those layers run together, the output is emergent. The behavior comes from the specific collision at that moment in time. And this changes the scope of the question enterprises can ask.

A REIMAGINING OF THE MARKET RESEARCH PARADIGM

The global market research sector operates as a $153 billion annual subscription to the past. Characteristically, more than 60% of this capital is directed toward human service providers rather than data networks or platform tools. This conventional approach forces brands to procure periodic insights at isolated stages of a project cycle simply to address narrow, predefined inquiries.

We believe a fundamentally distinct approach is needed: continuous infrastructure. Instead of serving as a backward-looking validation tool for pre-determined choices, we can establish an active ecosystem where target audiences remain perpetually accessible during the decision-making process.

For one client, this approach drove a 10x surge in consumer engagement frequency through AI agents compared to traditional human feedback. The process accelerated innovation timelines by 65% and secured $7.5 million in total financial savings.

We envision a future in which all marketing executives have access to an always-on operational engine. When they can co-create novel ideas, evaluate concepts in real time, and simulate complex market launches, the avenues for innovation and success multiply.

A major shift in expectations is underway, as enterprises now demand to understand both their customers’ identities and their behavior in specific scenarios. Rather than disappearing, the concept of the persona is expanding, and succeeding in this new landscape requires a powerful technological foundation.

Although we have always designed simulations, we finally have the right tools to construct them properly.

Chad Reynolds is the founder and CEO of Vurvey Labs