Research question
How often do estate or probate documents contain review errors, and how often do those errors create rework? The short answer is that no representative U.S. estate-document study located for this article supports a national error-rate statistic. There is, however, a useful probate-specific public proxy and newer legal-document review research that shows why scope and definitions matter.
The evidence in one table
| Evidence | Reported result | What it measures | What it does not establish |
|---|---|---|---|
| Ireland Courts Service annual report, 2020 | Error rate **over 70%** on first-time probate applications; **25%** on applications submitted a second time | Errors and rework in probate applications submitted to Ireland’s Dublin Probate Office and District Probate Registries | A U.S. rate, an estate-contract rate, or the probability that a particular document is wrong |
| Ireland Courts Service annual report, 2020 | Almost **30,000 applications** had to be checked in practice to issue the reported grants | Operational checking workload after rework | A clean national denominator for all estate documents |
| Keeling, Mangum, Hanke, and Ogden, *Journal of Technology Law & Policy*, 2026 | **83.9% recall** and **84.7% precision** in a 1,600-document GenAI relevance-review study | Classification performance in one real-world legal/regulatory document set | Human estate-review error, probate rework, or a recommendation for unsupervised use |
| Representative U.S. estate-document review benchmark | **Not established in the sources reviewed** | Evidence gap | That the true rate is zero or that all review teams perform alike |
The Irish figures come from an official court-service report, which says the published grant totals do not account for errors on submitted papers. Its error rate is therefore best understood as a workflow rework signal. The legal-tech study is peer-reviewed research, but its recall and precision measures answer a different question.
What the probate rework proxy tells us
The Courts Service Annual Report 2020 describes the Irish probate process and reports an error rate above 70% on first-time applications, falling to 25% on applications submitted for a second time. It says the Dublin Probate Office and District Probate Registries had to check almost twice the number of applications shown in the grant totals and, in practice, almost 30,000 applications in 2020.
That is directly relevant to the idea of rework: an application can return to the queue after a problem is found. But the statistic is not a measure of every field, clause, signature, deed, will, trust, or estate contract. It is also from Ireland, in 2020, and reflects a court-submission workflow. It should be used as a clearly labeled proxy, not copied into a U.S. marketing claim.
What legal-document research adds
The peer-reviewed article *Inconsistent Responsiveness Determination in Document Review: Difference of Opinion or Human Error?* examined inconsistent assessments of the same document collection using the TREC 2009 Legal Track. The authors concluded that inconsistency was due in large part to human error. The study supports a practical point: review quality can be tested against a defined collection and a defined decision rule. It does not supply an estate-document percentage.
A newer peer-reviewed study, *Prompting Change with GenAI in Large-Scale Document Review: A Real-World Study*, compared human coding with Relativity’s aiR for Review on a 1,600-document sample. Its reported 83.9% recall and 84.7% precision are classification results. Recall is not the inverse of an estate-document error rate, and precision is not the share of an estate file that is accurate. The study itself reports variation by document type and mixed issue-coding results.
Relevance to estate contracts and transaction risk
For an estate contract, a missed or unresolved discrepancy can affect more than wording. Names, authority, dates, legal descriptions, signatures, beneficiary or fiduciary roles, tax provisions, and attachments may need to agree across multiple records. A review log should show which source was checked and whether the issue was resolved, escalated, or left open for a qualified professional.
Teams can pair this evidence with a plain-language explanation of how to read an estate contract and a clause-focused review of estate contract clauses. The companion research on estate document review error rates explains why IRS estate-tax filing counts cannot serve as a document-review denominator.
The operational goal is not to promise a universal “error-free” review. It is to reduce avoidable rework, make exceptions visible, and preserve an audit trail for issues that require attorney, tax, title, or fiduciary judgment.
A practical measurement model
Before publishing a rate, define the unit and denominator. These measures should not be blended:
| Measure | Example denominator | Useful companion fields |
|---|---|---|
| First-pass rework rate | Applications or matters returned for correction | Jurisdiction, submission type, reason, resubmission date |
| Document discrepancy rate | Documents with a confirmed discrepancy divided by documents checked | Document type, version, source record, severity |
| Check failure rate | Failed checks divided by checks performed | Check definition, reviewer, second-review result |
| Matter exception rate | Matters with at least one confirmed exception divided by matters reviewed | Inclusion rules, review period, disposition |
Keep “flagged,” “confirmed,” and “corrected” separate. A reviewer’s question is not automatically an error, and a corrected item is not necessarily evidence of who caused it. If a second reviewer changes a disposition, retain both decisions and the reason for the change.
Methodology and scope note
Research completed **August 5, 2026**. The review prioritized an official court or agency report with a stated probate error measure and peer-reviewed legal-document research with transparent sample and performance definitions. The sources were read for their stated populations and metrics; no estate-specific rate was inferred from a general legal-tech result. The Irish report is a jurisdictional and temporal proxy, not a U.S. benchmark. The two legal-document studies are qualitative or classification-performance evidence, not estate-contract error studies. This article does not claim that a national rate exists, that the true rate is zero, or that any cited result predicts an individual matter.
Frequently asked questions
What is the estate document review error rate?
No representative U.S. estate-document study reviewed here establishes a national rate. The closest directly relevant public proxy is the Irish Courts Service result: above 70% on first submissions and 25% on second submissions in 2020.
Does the 70% probate figure apply to U.S. estate contracts?
No. It measures Irish probate applications, not U.S. estate contracts. It shows that rework can be substantial in a probate filing workflow, but it cannot predict a particular transaction.
What do recall and precision mean in document review?
Recall measures how much of the relevant material a system identifies. Precision measures how much of what it identifies as relevant is actually relevant. Neither is an estate-document error rate.
How should an estate team measure review quality?
Define the unit, population, period, discrepancy taxonomy, reviewer process, and denominator. Track first-pass rework separately from confirmed discrepancies, and preserve the evidence supporting each disposition.
Can an estate contract review service guarantee a lower error rate?
No responsible service should promise a universal rate without a defined population and audit method. A workflow can make source checking and exception handling more consistent, but legal and tax judgments still require qualified professionals.
Sources
- Courts Service, Ireland, *Courts Service Annual Report 2020*, probate section, pp. 73-74.
- Maura R. Grossman and Gordon V. Cormack, “Inconsistent Responsiveness Determination in Document Review: Difference of Opinion or Human Error?”, *Pace Law Review* 32 (2012), 267.
- Robert Keeling, Ray Mangum, Amy Hanke, and Alyssa Ogden, “Prompting Change with GenAI in Large-Scale Document Review: A Real-World Study”, *Journal of Technology Law & Policy* 30 (2026).
**Professional disclaimer:** This article is general research, not legal, tax, accounting, fiduciary, financial, or quality-assurance advice. It does not establish an attorney-client relationship. Estate documents, probate procedures, transaction requirements, and review standards vary by jurisdiction and facts. Have a qualified professional confirm the applicable requirements before relying on a statistic or workflow decision.