mlsanityData to 7 Oct 2026

Persona Conditioning of Brand Recommendations in Retrieval-Augmented Commercial Chat: A Prominence-Stratified Cross-Provider Audit

arXiv 2605.30207 ↗First posted 28 May 2026cs.AIcited by 1

On the map: Retrieval-augmented generation → RAG pipeline design and evaluation

Abstract (arXiv)

The same prompt -- "best CRM software" -- reaches AI assistants from buyers in widely different contexts: a solo founder, an enterprise VP, a UK SMB owner. We audit how strongly that contextual variation reshapes which brands the model recommends. The audit samples 2,000 runs over a design space of 10 personas x 8 prompts x 3 model configurations x N=10 reps, with the two OpenAI cells at full 8-prompt coverage and the Anthropic sonnet-4.6 / low cell at 4-prompt coverage. Prefixing the user message with a persona drops the recommendation-set similarity (Jaccard) by Delta = -0.12 to -0.20 relative to a same-persona baseline (clustered 95% CIs exclude zero on all three measured cells; the sonnet cell's CI rests on only 4 prompt clusters and is correspondingly wider). The effect is sharply prominence-stratified: category leaders are persona-resistant (~80% same-brand consistency across personas), but mid-market brands swap up to 75% of the recommendation set as the persona changes. The Anthropic model shows a larger point-estimate effect than the OpenAI configurations, though clustered CIs overlap for the closer contrast (sonnet vs. OpenAI/high); the asymmetry is consistent with Anthropic's more retrieval-unattributed generation route (43-52% recommendations without observed retrieval-layer evidence, vs OpenAI's 8-29%, documented in Jack 2026). Any measurement of AI brand perception must condition on the buyer persona supplying the query: the same prompt produces materially different recommendation sets depending on who the model thinks is asking, and a measurement protocol that aggregates across personas systematically obscures that variation. The effect concentrates at mid-market and is largest on the most priors-reliant generation route in our audit, consistent with persona responsiveness growing as models lean more on training-data priors and richer context integration.

Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Placed in RAG pipeline design and evaluation, one of 15 directions in Retrieval-augmented generation. Builds on 23 of its 26 references that are in the map. First posted on arXiv 28 May 2026.

Builds on

23 of 26 references.

References · 23

  1. 2022Emergent Abilities of Large Language Modelscited by 2,483
  2. 2022When Not to Trust Language Models: Investigating Effectiveness of Parametric and Non-Parametric Memoriescited by 1,103
  3. 2023Quantifying Language Models' Sensitivity to Spurious Features in Prompt Design or: How I learned to start worrying about prompt formattingcited by 778
  4. 2024Scaling Synthetic Data Creation with 1,000,000,000 Personascited by 397
  5. 2023State of What Art? A Call for Multi-Prompt LLM Evaluationcited by 299
  6. 2024Knowledge Conflicts for LLMs: A Surveycited by 274
  7. 2024Quantifying the Persona Effect in LLM Simulationscited by 164
  8. 2024POSIX: A Prompt Sensitivity Index For Large Language Modelscited by 56
  9. 2023GEO: Generative Engine Optimizationcited by 53
  10. 2023When "A Helpful Assistant" Is Not Really Helpful: Personas in System Prompts Do Not Improve Performances of Large Language Modelscited by 50
  11. 2025Exploring the Impact of Temperature on Large Language Models:Hot or Cold?cited by 40
  12. 2025The Prompt Makes the Person(a): A Systematic Evaluation of Sociodemographic Persona Prompting for Large Language Modelscited by 40
11 more
  1. 2024Stereotype or Personalization? User Identity Biases Chatbot Recommendationscited by 33
  2. 2024Large Language Models as Recommender Systems: A Study of Popularity Biascited by 25
  3. 2025Population-Aligned Persona Generation for LLM-based Social Simulationcited by 20
  4. 2024LLM Whisperer: An Inconspicuous Attack to Bias LLM Responsescited by 15
  5. 2024Context-Parametric Inversion: Why Instruction Finetuning Can Worsen Context Reliancecited by 9
  6. 2025Unveiling Knowledge Utilization Mechanisms in LLM-based Retrieval-Augmented Generationcited by 4
  7. 2026Auditing Preferences for Brands and Cultures in LLMscited by 4
  8. 2026Paraphrase Brittleness in Production Retrieval-Augmented Commercial Recommendation: Reproducibility Below the Rerun-Stability Baselinecited by 2
  9. 2026Gender and Race Bias in Consumer Product Recommendations by Large Language Modelscited by 1
  10. 2026Prominence-Stratified Failure Modes in Retrieval-Augmented Commercial Recommendation: A 37,000-Run Auditcited by 1
  11. 2026Divergent Recommendations, Convergent Diagnoses: Cross-Provider Failure-Mode Convergence in AI Commercial Recommendationcited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026"If I Had to Buy Just ONE: Galaxy S26 Ultra": Auditing AI-Generated Product Recommendationscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2026Per-Entity Bias Mapping for AI Visibility: Why Brand Mentions Require Entity-Specific Calibration
  2. 2025PersonaBOT: Bringing Customer Personas to Life with LLMs and RAG
  3. 2026Benchmarking the Personalization Capabilities of Large Language Models
  4. 2026When Persona Attributes Improve Population Alignment in Large Language Models
  5. 2026Your Language Model Secretly Contains Personality Subnetworks
  6. 2025Is Active Persona Inference Necessary for Aligning Small Models to Personal Preferences?
  7. 2024PersonalLLM: Tailoring LLMs to Individual Preferences
  8. 2026Whose doctor does the AI recommend? An algorithm audit of reputation and demographic signals in large language model-assisted physician choice