- Key insight: Find out how specific smaller CPA firms are outperforming giant competitors in AI search.
- What's at stake: Busiest accounting practices risk losing client leads if AI bots continue ignoring their services.
- Supporting data: 87.2% of AI search results were concentrated within one group of firms.
Recent data from two different studies has found that the majority of CPA firms are completely invisible to AI search, which is dominated by a small number of extremely large organizations, but the exceptions show that size alone is not the sole factor in determining whether or not a bot mentions an accounting practice.
In a study from marketing advisory firm
What they found was that 25 firms account for 87.2% of all AI search results: RSM, BDO, Baker Tilly, CLA, Grant Thornton, Forvis Mazars, CohnReznick, EisnerAmper, Cherry Bekaert, Crowe, Plante Moran, Withum, CBIZ, Armanino, Wipfli, Aprio, Citrin Cooperman, Eide Bailly, PKF O'Connor Davies, BPM, BerryDunn, Sikich, LBMC, the Bonadio Group and Doeren Mayhew. The first 10 of these firms accounted for 61.3% of all mentions. The No. 1 most mentioned firm was RSM, which was mentioned 29.48% of the time. The rest of the Top 100 Firms struggled to be recognized: 79 of the Top 100 were mentioned only 3% of the time. Eight firms were never mentioned at all.
Overall, there is a strong relationship between firm size and AI search visibility. When the Top 100 Firms are divided into five tiers based on revenue, median firm appearance rates fall sharply with each successive tier. The median firm in the highest revenue tier appeared in AI responses roughly 19 times as often as the median firm in the third tier, and more than 350 times as often as the median firm in the lowest tier.
However, size might be more of a correlation versus a causation, as the study also found that several smaller firms substantially outperformed what their revenue rankings would predict. Larger firms, because they have more market presence, naturally generate more artifacts pointing to them as an authority on something and therefore more likely to show up in AI recommendations. However, smaller firms can get outsize results through an established association with a particular industry, service, locale or even a type of problem that they're known especially well for addressing. And so it seems that what is more important is a large digital footprint that establishes a firm as an authority in a particular area.
However even if a firm is highly competent in a particular area, this is not a guarantee that AI will consistently recommend them. This was one of the major findings in another recent study conducted by
Their study looked specifically at whether AI assistants recommend the CPA firms that actually perform single audits in the Research Triangle, North Carolina, and on what evidence. They chose single audits because any organization spending $1 million or more in federal funds in a year must have one, and every one is filed publicly with the Federal Audit Clearinghouse, which names the auditor. It is the rare professional service with a complete public record of who does the work. The study involved 10 Triangle-area firms named as auditor on three or more single audits in the federal record (AY2023–2025).
Researchers developed 10 buyer questions derived from the kinds of organizations the panel firms actually audit. The questions were run through five AI models (ChatGPT, Claude, Perplexity, Gemini and Google AI Mode) five times each through their public developer interfaces with no chat history and no personalization. This produced a sample of 250 answers on Aug. 11, 2026, and 250 on Sept. 28, 2026, with the same model versions both times. A firm counts as recommended when its name (or a listed former name) appears in the answer text. Every count was hand-verified against the raw answers before publication. The researchers looked at the results in the context of 10 firms that served as a panel.
What they found was that the firms doing this work the most were recommended the least. The four busiest firms in the panel, representing 45 of 75 federal engagements in the area, were recommended 24 times in August. Conversely, the six least busiest firms, representing 30 engagements between them, were recommended 184 times. The firm named most often is one of the panel's three smallest filers. In contrast, the single most prolific nonprofit auditor in the entire record was named exactly zero times in the entire experiment despite having more single audit clients than any firm in the panel.
Like the other study, this one found digital footprint mattered, but only in a broad manner. Overall, the most frequent thing bots referenced when making recommendations was the firm's own websites, citing them more than any other source. Website alone, however, was not a decisive factor. Firm websites that were heavily cited tended to name their specific geographic region (i.e., the Triangle), describe the buyer's specific kind of work at length and had been updated recently. The study noted that the busiest firm in the panel had 1,821 pages but did not mention place in any of them. Beyond firm websites, bots were also likely to cite government and client documents mentioning the firm, industry associations and organizations, and commercial directories and ratings databases.
In contrast, bots tended to pass over sites like Reddit, Quora, Wikipedia and other large public sites. Further, while they also search social media, that generally just means LinkedIn company and profile pages. They seldom referenced new coverage, usually only when there was a big list or ranking. Further, despite citing 2,725 web pages across the experiment, the Federal Audit Clearinghouse was cited only four times and the North Carolina State Board of CPA Examiners was never cited at all.
However, certain confounding factors make it difficult to fully extrapolate the findings into a consistent method. For example, the second most named firm, mentioned 78 times, had no website at all. The assistants learned its name from a charter school's archive of old audit reports (which shows the firm stopped being that school's auditor three years ago), the State Treasurer's files and a town council's minutes.
AI is
Introductory bullet points created by AI with editorial review.
