Contextual Data Improves AI Brand Perception, Study Finds

Image credit: Search Engine Journal

Providing artificial intelligence models with comprehensive contextual information can dramatically improve their perception of a brand, shifting recommendations from negative warnings to positive endorsements, a recent experiment found.

AI systems frequently over-correct for potential recommendation risks by highlighting virtually all complaints or poor reviews, even minor or unsubstantiated ones, which often leads to an unfavorable portrayal of brands.

Frontier AI systems from companies like OpenAI, Anthropic and Google are specifically instructed to exercise caution to prevent harm, particularly in areas concerning personal finance or health, known as Your Money or Your Life (YMYL) topics.

AnswerShare, a company specializing in AI content strategies, conducted an experiment demonstrating that furnishing full context — including total customers, years in business, company responses to feedback and a complete business description — significantly enhanced AI recommendations for one of its clients.

Paul Harvey, an expert in AI reputation management, explained that this method allows AI models to understand the scale and operating history of a business, rather than focusing solely on isolated negative feedback.

Within 14 days of providing this comprehensive brand context, AI recommendations for the client surged from 23.8 percent to 100 percent, AnswerShare reported. While complaints were still discussed, they were presented alongside the company’s overall operational scale and history.

The methodology involves deploying ‘workers’ that route web traffic to both machine learning crawlers and human verification layers. Contextual content is then strategically placed on these workers at the content delivery network (CDN) edge, ensuring it is readily accessible to AI systems.

This approach aims to counteract the AI’s default behavior of surfacing almost every complaint or poor review, a tendency that can unfairly damage a brand’s online reputation.

The experiment suggests that by offering a more balanced and complete picture, businesses can influence how large language models and other AI systems interpret and present information about them to users.

Source: Search Engine Journal