AI Assistants Falter in Recommending Specialized Accounting Firms

The report, titled "Three Doors and a Brick Wall: How AI Assistants Choose CPA Firms," tested five major AI platforms across 250 queries. Researchers compared these outputs against Federal Audit Clearinghouse data regarding Single Audits—mandatory financial reviews for organizations receiving over $1 million in federal funding. The findings highlight a disconnect between machine recommendations and actual professional competence. The four firms handling the majority of regional audit volume appeared in results only 24 times, while smaller, less experienced firms were suggested 184 times. Three high-volume firms were ignored entirely.

AI models appear to favor firms that actively maintain specific website content, such as targeted regional keywords and lengthy descriptions of client types. This strategy often bypasses authoritative sources; the Federal Audit Clearinghouse was cited just four times across 2,725 source references. The reliability of these suggestions is further compromised by data staleness: in dozens of instances, assistants recommended firms under names retired years ago or directed users to websites currently under construction. Heather Harreld, founder of AIWorthy, noted that businesses are currently investing in AI visibility without understanding how these models actually process information. According to Harreld, firms need to see the specific path a machine takes to reach a recommendation rather than relying on vanity metrics.