340-Response Audit Proves Legacy Sentiment Forces AI Search to Recommend Closed Businesses

Field tests reveal that AI search engines exhibit an entity invalidation lag, continuing to recommend defunct brands years after their physical closure.

Analyzing 340 prompt responses across four AI systems, the data proves that unrefuted legacy sentiment inside training corpora overrides factual operating status, observes Sebastian Prohaska.

This retrieval flaw sustains zombie entity visibility across LLMs.

Google AI actively recommends a 212-location chain that closed in December 2023.

Similarly, the defunct brand Fit One retains a 3% recommendation visibility.

The algorithm prioritizes historical praise over current reality; AI systems actively recommend the closed chain High5 strictly because legacy content highlights its "fairest cancellation rules."

Separately, the audit confirms that physical footprint fails to dictate AI visibility.

A 14-location brand captures a 10% visibility share, while a 160-location competitor achieves only 6.2%.

Brand dominance also fractures across models; the top market leader secures an 87.4% recommendation rate in Google AI but drops to 29% in Perplexity.

Neutralizing this invalidation lag requires consulting GEO experts rather than manual prompt manipulation.