IRS Sent 20,000 Ghost-Preparer Letters, but Its Data Still Cannot Connect the Full Enforcement Chain

A measurable mailing is not yet a measurable enforcement result

IRS mailed 20,000 letters in a 2026 pilot aimed at taxpayers whose returns might have been prepared by undisclosed paid preparers. By May, fifty-two recipients had submitted preparer complaints, according to a federal audit released October 16. That is 0.26% of the mailing, calculated by What’s the Scoop With Broach—but it is a dated complaint-response measure, not a fraud-detection rate or a final assessment of the pilot.

The more consequential finding appears when that activity is matched with the audit’s data-system analysis: IRS could count selected mailings and responses, but its systems did not provide a connected account of how many ghost preparers it had identified, investigated and addressed. A complaint, a corroborated potential lead, a compliance closure and a criminal outcome are different units.

We reconstructed those distinctions from the report’s pilot description, system limitations, enforcement section and IRS’s written response. The accountability question is whether a taxpayer-protection initiative can show what happened after it generated a lead, rather than treating activity volume as proof of protection.

The earlier pilot supplies a different denominator

For its 2025 pilot, IRS recorded more than 2,000 responses, generating more than 200 distinct ghost-preparer leads. Sixty-eight were corroborated as potential ghost preparers. Multiple taxpayers sometimes supplied information about the same preparer, so responses cannot be counted as distinct preparers.

The report does not supply a comparable mailing denominator and final outcome chain for that 2025 pilot. Its 2026 count is also a May snapshot and used a different response route: an electronic complaint form rather than the earlier telephone process. Dividing sixty-eight by the later year’s 20,000 letters, or declaring the two pilots improved or deteriorated on these figures alone, would manufacture a comparison the records do not support.

The analytical distinction also protects taxpayers from being mislabeled. A return flagged as possibly involving an undisclosed preparer is not proof that its filer committed fraud, and a corroborated potential preparer lead is not a conviction. The report says the pilot both educated taxpayers and sought identifying information.

Where the handoffs lose visibility

The Return Preparer Office’s Ghost Inventory Tracking System records cases and compliance actions, but does not track outcomes after referrals to other offices. Some preparers are brought into compliance before a case is opened, so its case count is not a census of people identified.

The Small Business/Self-Employed division’s Return Preparer Database lacks a flag showing whether a preparer was referred as a ghost preparer and later brought into compliance. It also did not separately track the relevant failure-to-sign and preparer-identification-number penalty subsections until 2025. Earlier grouped penalty counts therefore cannot be treated as a clean series of ghost-preparer sanctions.

Criminal Investigation’s management system lacks a unique ghost-preparer fraud-scheme code. Auditors found its manual search incomplete. Separately, the research office shared analytical leads but did not know which offices used the information or whether it resulted in confirmed cases or enforcement. These are distinct handoff problems; a new dashboard displaying the existing counts would not by itself resolve them.

Existing results and agency repairs deserve the same precision

The audit documents genuine compliance work. From 2022 through 2025, the Return Preparer Office opened around 1,400 cases annually and closed about 60% because preparers returned to compliance after a letter or telephone contact. That supports an administrative compliance result for those cases. It does not establish that 60% of all hidden preparers were detected or that all possible fraud was resolved.

The 2025 return-preparer office referral count was forty-three cases to other business units. That is not the total number of criminal investigations across IRS: other routes can generate cases, and the systems do not provide a comprehensive criminal count. The report’s forty-four illustrative federal court cases were selected from publicly identified cases and agency records, not a statistical national sample.

Counterevidence also appears in the education review. GAO rated IRS consistent with seven of nine consumer-education planning practices and partially consistent with two: audience research and success metrics. The documented shortcoming is not an absence of public warnings. Our fresh check of IRS’s fraud-reporting page confirmed a dedicated route for reporting tax-return-preparer misconduct.

In its September 29 response, IRS agreed to all three recommendations: review data quality and sharing, continue developing a coordinated preparer strategy, and define measurable objectives and performance measures. The response commits to further work; it does not state that integrated outcome tracking is already complete. The official recommendation page listed all three open when checked October 17.

This report is original document analysis of the pilot denominators, the referral and tracking gaps, and the difference between compliance outputs and final outcomes. We checked the arithmetic, read the audit’s methods and substantive findings, and rendered the written agency response for examination. Billy Roach did not contact taxpayers, inspect tax returns or conduct interviews. The public record supports demanding connected outcome evidence; it does not support inventing a nationwide fraud total or calling every unanswered letter a failed investigation.