# Do Synthetic Personas Predict Real Audience Response? A Sim-to-Real Study Where a No-Persona Baseline Beats Persona-Based Copy Simulation

Published: 2026-09-23 · Updated: 2026-09-23 · Category: Research

Source page: https://www.sipply.dev/story/do-synthetic-personas-predict-real-audience-response-a-sim-28babdb8 · 中文: https://www.sipply.dev/zh/story/do-synthetic-personas-predict-real-audience-response-a-sim-28babdb8

## Summary

A recent study published on arXiv examined the effectiveness of synthetic personas created by large language models (LLMs) in predicting audience responses to marketing copy. The research found that a baseline approach, which did not utilize any personas, outperformed the persona-based simulations. This suggests that the incorporation of synthetic personas may introduce biases that ultimately hinder the accuracy of predictions regarding audience engagement and response. The implications of these findings challenge the conventional reliance on personas in marketing strategies.

## Key takeaways

- No-persona baseline strategies may yield better predictive outcomes than persona-based approaches.
- Synthetic personas can introduce bias, potentially skewing audience response predictions.
- The study raises questions about the effectiveness of traditional marketing methods that rely on personas.

## Why it matters

This research highlights the need for marketers to critically evaluate the tools and methods they use to understand their audiences, particularly the reliance on synthetic personas that may not accurately reflect real-world responses.

- **Product managers:** For product managers, these findings suggest a reevaluation of user persona strategies in product development and marketing, emphasizing the importance of data-driven approaches over potentially biased simulations.
- **Developers:** Developers and engineers should consider the implications of using LLMs for generating personas, as the study indicates that simpler, more direct methods may yield more reliable results in audience engagement predictions.
- **Students and job seekers:** For students and job-seekers, this research underscores the importance of understanding audience dynamics and the potential pitfalls of relying on synthetic models, encouraging a focus on data integrity and real-world testing in marketing strategies.

## Sources

- [Do Synthetic Personas Predict Real Audience Response? A Sim-to-Real Study Where a No-Persona Baseline Beats Persona-Based Copy Simulation](https://arxiv.org/abs/2609.25010) (arXiv cs.AI, 2026-09-23)

## 中文

### 合成角色能否预测真实受众反应？一项无角色基线优于基于角色的复制模拟的模拟与现实研究

最近在arXiv上发表的一项研究考察了大型语言模型（LLMs）创建的合成角色在预测受众对营销文案反应方面的有效性。研究发现，未使用任何角色的基线方法优于基于角色的模拟。这表明，合成角色的引入可能会导致偏见，从而妨碍对受众参与和反应的准确预测。这些发现的含义挑战了传统营销策略中对角色的依赖。

**要点：**

- 无角色基线策略可能比基于角色的方法产生更好的预测结果。
- 合成角色可能引入偏见，潜在地扭曲受众反应预测。
- 该研究对依赖角色的传统营销方法的有效性提出了质疑。

**为什么重要：** 这项研究强调了营销人员需要批判性地评估他们用来理解受众的工具和方法，特别是对可能无法准确反映现实反应的合成角色的依赖。

---
Sipply: Sip AI News Simply. https://www.sipply.dev