PERCEPTION
The AI's Vibe Check: Understanding Perception in Bonafide
In the world of generative AI, customers are increasingly asking Large Language Models (LLMs) not just for facts, but for opinions and recommendations about your brand,. To succeed, your brand needs to know how the AI genuinely perceives you—which is precisely what the Perception metric in Bonafide measures.
|
Perception for Destinations |
Perception for Hotels
|
What is Perception?
Perception captures the generated preferences and opinions that Large Language Models (LLMs) have developed about your brand. It is essentially the AI’s synthesized understanding of the general public’s opinions and feelings. Unlike Accuracy, which focuses on verifiable facts, Perception assesses the personality and subjective reputation of your brand.
LLMs form these nuanced opinions by thoroughly scraping vast quantities of text from the web, including traveler reviews and other subjective content. This process gives the AI a broad understanding of how people talk about your brand in the digital landscape.

The Elements of Brand Perception
Perception assesses a comprehensive list of aspects that contribute to your brand’s overall image, ensuring the AI understands the complete picture.
The Risk of Skewed Perception
A primary challenge with how LLMs form opinions is their wide, often indiscriminate, data consumption. LLMs might pick up random, isolated bits of information, such as one-off negative comments on platforms like TripAdvisor or specific negative reviews buried on Reddit. This isolated data can potentially skew the AI’s overall perception of your brand’s image, creating an unverified and potentially harmful narrative. Perception scores are relative and often measured in comparison against your brand's peers.
Controlling the Narrative
When a perception issue is identified, the Bonafide platform provides a critical mechanism for your brand to control the narrative and shape the AI's understanding.
Instead of hoping the LLMs correct themselves (which is unlikely), you proactively fix perception by feeding real, verified data directly into the LLMs. This process is crucial and involves:
-
Ensuring Factual Content: Making sure your website contains correct, robust facts.
-
Optimizing Visuals: Providing high-quality photos paired with clear, descriptive captions that LLMs can easily crawl and understand.
By becoming the undeniable primary authoritative source of truth about your brand, you directly influence the information LLMs use to form and cite their perceptions.
What's Being Judged?
Perception assesses various aspects of a brand and vary by the type. Some example industry types and perception measures:
|
HOTELS |
Destinations Marketing Organization (DMO) |
AIRLINES |
CRUISE |
|---|---|---|---|
|
Overall Experience
|
Airport & In-Flight Experience Booking & Pre-Travel Experience Brand Reputation Competitive Comparison Customer Control & Customization Ease of Use/Simplicity Fairness & Equity Loyalty Program Perception Transparency of Pricing & Fees Trust and Intent Unbundling Services Value for Money |
|

How LLMs Form Perceptions?
The LLMs determine these perceptions by "scraping the web". This means they gather information from various online sources. However, this can sometimes lead to issues because the LLMs might pick up "random things" like negative reviews from places like Reddit or random Trip Advisor comment, which could impact a BRAND's perceived image.
How Does Bonafide Help Improve Perception?
If a BRAND has a "perception issue," Bonafide aims to fix this by helping the BRAND control the narrative:
-
It's about distributing the "real true data" directly from the BRAND's brand to the LLMs.
-
This includes ensuring the BRAND's website has correct information and high-quality pictures with clear captions that LLMs can easily understand and "crawl" (read).
-
The goal is for the BRAND's own website and verified information to become the "source of truth" that LLMs cite, rather than less reliable sources like social media or third-party review sites.
-
By becoming the primary, verified source, the BRAND can influence what information LLMs present, which in turn helps change the perception. It's not about providing "high reviews" to Bonafide, but about making the BRAND's own data the definitive source for the LLMs.
-
This process is part of a larger "orchestration" solution, which involves multiple steps to improve the visibility and accuracy of the brand’s data for LLMs.


Strategic Goal of Perception Management
The ultimate goal of using the Curation process to manage Perception is to ensure the BRAND's own website and verified content becomes the primary authoritative source of truth that LLMs cite.
-
By correcting errors and verifying answers, the TRAVEL BRAND directly influences the information LLMs use to form their perceptions.
-
Bonafide aids this by creating special data packages (FAQ-style Q&As of over 500 data points) that LLMs easily consume.
-
Bonafide then uses its orchestration solution—improving website data, distributing data via partnerships like the Common Crawl Foundation, and direct distribution to companies like OpenAI, Microsoft, and Google—to ensure the verified content reaches the AI ecosystem. This process is turnkey, removing the manual work of distribution for the BRAND.
In Summary: Imagine Perception as a cumulative online report card given by the AI. If the AI is studying random gossip and old, unverified blog posts (subjective content scraped from the web), its opinion of your brand might be negative or inaccurate. By actively deploying verified knowledge using tools like Bonafide, you ensure the AI's "textbook" is your brand's own verified content, guaranteeing that its recommendations and synthesized opinions align with the story you want to tell.
APPENDIX
I. Key Terminology for Brand Perception Management
|
Term |
Detailed Definition |
|---|---|
|
Perception |
This metric captures the generated preferences and opinions that Large Language Models (LLMs) have developed about a BRAND. This synthesized understanding is formed by analyzing traveler reviews and other subjective content scraped from the web. |
|
Perception Assessment Aspects |
Perception assesses a comprehensive list of attributes related to the BRAND’s personality and character. These aspects include: accommodations, attractions, experience and highlights, budget and affordability, character and vibe, cleanliness, condition (how well-maintained the property is), location, services offered, value for money, climate and weather, food, health and safety, sustainability and environment, transportation, and the reputation of the BRAND itself. |
|
LLM Hallucination/Skew |
This issue arises when LLMs scrape vast quantities of text, potentially picking up random negative reviews buried on sources like Reddit or one-off comments on Trip Advisor. These isolated bits of subjective content can skew the AI’s overall perception of the BRAND's image. |
|
System of Record (aka Source of Truth) |
The BRAND’s own website, CMS, or PMS data. The goal is for the BRAND’s own website and verified information to become the primary authoritative source of truth that LLMs rely on and cite, thereby influencing Perception. |