Accuracy

ATR™ is the reputational risk metric of the FAIV™ framework. Unlike visibility metrics that measure presence, ATR™ measures information quality: how reliable is what AI says about you when someone asks.

It is built through a systematic verification process: claims that models make about the brand are collected (founding year, number of employees, services, locations, clients, awards, certifications) and cross-referenced with verifiable company data. Each claim is classified as correct, partially incorrect, incorrect or fabricated.

A low ATR™ is not just an image problem — it is a business problem. A potential client who asks AI about your company and receives incorrect information may discard you before contacting you, or contact you with incorrect expectations that generate friction in the sales process.

Where ATR™ evaluates the overall reliability of information, FAI™ quantifies it with precision. Against a defined set of verifiable claims (variable number depending on the sector and company size, typically 40-80 verification points), FAI™ measures exactly what percentage of what AI states about your brand is correct.

The methodology distinguishes three types of error: factual error (incorrect but existing in some source), pure hallucination (invented data with no identifiable source) and attribution error (correct data from another company attributed to yours). Each type has different implications for the correction strategy.

FAI™ is a Research-stage metric, meaning the corpus and verification methodology is under refinement. Early pilot studies show that even brands with high AI visibility have FAI™ between 60-75%, with errors concentrated in numerical data (employees, revenue, number of clients) and in offer attributes.

Unlike FAI™ which measures all types of inaccuracy, AHR™ focuses exclusively on pure hallucinations: claims that AI models generate about your brand that have no identifiable origin in any real source. They are not interpretation errors or attribution errors — they are invented data.

Hallucinations are especially prevalent in mid-sized brands that have enough presence for AI to attempt a description but insufficient structured content for models to be accurate. The model 'fills in the gaps' with plausible but false data.

AHR™ is a Research metric because the rigorous distinction between pure hallucination and factual error requires an exhaustive verification methodology that is under refinement. The most severe cases of high AHR™ involve legal risks if AI claims facts that could constitute false material information about the company.