LLM-Generated Ads: From Personalization Parity to Persuasion Superiority

TL;DR

The paper investigates LLMs as generators of persuasive advertising across personalization and universal persuasion. Through two controlled studies, it shows that AI can match human performance in trait-based personalization (best for openness) but exceed humans in delivering multimodal persuasion based on authority, consensus, and cognition. Notably, AI content remains superior even when participants correctly identify AI origin, highlighting a quality-driven advantage and a potential shift toward post-authenticity in advertising. The findings carry practical implications for scalable, low-cost creative production, while underscoring the need for ethical guidelines around manipulation, disclosure, and trust.

Abstract

As large language models (LLMs) become increasingly capable of generating persuasive content, understanding their effectiveness across different advertising strategies becomes critical. This paper presents a two-part investigation examining LLM-generated advertising through complementary lenses: (1) personality-based and (2) psychological persuasion principles. In our first study (n=400), we tested whether LLMs could generate personalized advertisements tailored to specific personality traits (openness and neuroticism) and how their performance compared to human experts. Results showed that LLM-generated ads achieved statistical parity with human-written ads (51.1% vs. 48.9%, p > 0.05), with no significant performance differences for matched personalities. Building on these insights, our second study (n=800) shifted focus from individual personalization to universal persuasion, testing LLM performance across four foundational psychological principles: authority, consensus, cognition, and scarcity. AI-generated ads significantly outperformed human-created content, achieving a 59.1% preference rate (vs. 40.9%, p < 0.001), with the strongest performance in authority (63.0%) and consensus (62.5%) appeals. Qualitative analysis revealed AI's advantage stems from crafting more sophisticated, aspirational messages and achieving superior visual-narrative coherence. Critically, this quality advantage proved robust: even after applying a 21.2 percentage point detection penalty when participants correctly identified AI-origin, AI ads still outperformed human ads, and 29.4% of participants chose AI content despite knowing its origin. These findings demonstrate LLMs' evolution from parity in personalization to superiority in persuasive storytelling, with significant implications for advertising practice given LLMs' near-zero marginal cost and time requirements compared to human experts.

Figures (8)

• Figure 1: The proposed generation pipeline. Stage 1 utilizes a KG to map user-product relationships, providing structured context for prompt construction. In Stage 2, the LLM transforms these constraints into persuasive narratives (Generative Storytelling), which are then delivered to matched audiences in Stage 3.
• Figure 2: Procedure of the experiment: (i) Task 1, including three questions presented on a 5-point Likert scale; (ii) Task 2, prompting participants to select one ad message from the four messages displayed side-by-side; followed by (iii) the Big5 questionnaire to gather their personality information. Note that there are four variants of Task 1: OH, OG, NH, and NG
• Figure 3: Example of advertisements used in Study 2 for the scarcity persuasion principle condition. Left: Human-created advertisement. Right: AI-generated advertisement (text generated using LLMs, image generated using a diffusion model - Midjourney). Both ads emphasize time-limited availability and urgency, core elements of the scarcity principle. Participants viewed both advertisements side-by-side without knowledge of which was AI-generated.
• Figure 4: Distribution of participant preferences (clicks) when ads were presented side-by-side in Task 2. LLM-generated ads collectively received 51.14% of clicks compared to 48.86% for human-written ads.
• Figure 5: Qualitative drivers of preference. Participants preferring AI (red) overwhelmingly cited Color/Tone and Model Quality, whereas those preferring humans (blue) prioritized Authenticity and Clarity.