State courts in the United States handle roughly 66 million cases a year. Each one is filed into a local system, under a local case-type code, by one of more than 200 separate reporting units that answer to no single national standard. Anyone using those records to decide on a lease, a loan or a policy inherits whatever that local code captured, and whatever it missed.
How much of the court docket actually speaks to risk?
The headline number is misleading by design. Traffic infractions alone accounted for about 46% of all incoming state court cases in 2024. Debt collection makes up roughly a quarter of civil cases and about 5% of the total docket.
Neither category tells you much about the counterparty in front of you. Both land in a records search the same way a felony does: as a hit, with a name attached.
The volume problem is real, but it is the cheap one. The expensive problem is that the records which do matter arrive describing themselves incorrectly.
Why do court records arrive misclassified?
In Brazil, a single national resolution standardizes case classes and subjects across the entire judiciary. The United States has no equivalent. Case-type codes are set county by county, state by state, and the same underlying conduct can carry different labels in two adjacent jurisdictions.
That produces four failure modes that any records-based decision inherits:
- The label is assigned at filing, not at resolution. A case is coded when the complaint is filed, before anyone knows what it becomes. A matter that opens as a collection action and ends as something else keeps its original code.
- Dispositions go missing. A record shows charges filed and never shows the dismissal. This is among the most common defects in background screening. It usually traces back to a vendor that never purchased the update, or to automated scraping that ignored what happened after the initial filing.
- One case looks like several. Arrest, charge and conviction are stages of the same proceeding. Pulled from multiple sources and presented as separate rows, they read as a pattern of offenses that never occurred.
- Nobody separates plaintiff from defendant. A judgment entered against someone is one signal. Three suits that person filed against companies in your sector is a completely different one. A raw records feed treats them identically.
Add the matching layer on top. The Consumer Financial Protection Bureau has stated that name-only matching does not satisfy the FCRA's accuracy standard. It remains common anyway, and it is how one person's record ends up attached to another person's file.
What does a misread court record cost under the FCRA?
This is where the U.S. calculation diverges sharply from most markets. A misclassified record is a bad decision and, on top of that, a documented liability.
Section 607(b) of the Fair Credit Reporting Act requires consumer reporting agencies to follow reasonable procedures to assure maximum possible accuracy. In January 2024, the CFPB issued an advisory opinion spelling out what that means for public records. Procedures must prevent reporting of expunged, sealed or otherwise restricted records. They must include existing disposition information when reporting arrests, charges, evictions or other filings. They must prevent duplicate entries, and they must present the stages of a single proceeding so they cannot be mistaken for separate cases.
Every one of those requirements maps directly onto a defect in the raw court record. The regulator described the default state of the data, not an exotic edge case.
The consequences are showing up in filings. FCRA litigation reached 6,053 lawsuits in the first nine months of 2025, up 30.7% year over year. The pace accelerated into 2026, with April filings running more than 60% above the same month a year earlier.
Meanwhile, the decision error runs in both directions. A routine civil matter coded into a generic criminal category becomes an alert. A good applicant is declined, or pushed into a manual review that takes days they will not wait through. A case that genuinely matters, filed under a vague catch-all code, passes as background noise and gets approved.
The first error costs conversion and creates dispute exposure. The second costs the loss.
How do you fix court record classification in a background check?
A checklist that applies to any operation touching court records, with or without Zarv:
- Separate plaintiff from defendant. If your report does not distinguish the two, it is not ready to decide on. It is the cheapest filter available and the most frequently skipped.
- Classify by content, not by code. Treat the county's case-type code as a hypothesis and the filing text as the evidence.
- Require the disposition. A charge without an outcome is an incomplete record. Under the CFPB's stated position, reporting it that way is a defect.
- Collapse duplicates before scoring. Stages of one proceeding gathered from several sources must resolve back to one case, or your model counts a single event three times.
- Weight for recency and repetition. A 2011 filing and seven filings of the same type in five years are different signals, whatever code they carry.
- Set depth proportional to the value at risk. A $40 transaction does not warrant the scrutiny of a $90,000 contract. When the operation declares what is at stake, friction stops being uniform.
- Monitor after approval. Correct classification today says nothing about the case filed in March. Verification without continuous monitoring is a snapshot.
Where does Zarv fit?
In the Zarv journey, this is the verification stage: deciding who to contract with in credit underwriting or insurance underwriting.
Zarv ID delivers court records that reflect what each case actually is, inside the search the operation already runs. Nothing changes in the workflow. What changes is the decision. Fewer good applicants are declined over records that never mattered. The cases that do carry weight stop hiding under a generic code.
The stage after it is monitoring, because an operation that only checks at the front door never sees the relationship change.
Frequently asked questions
What is a background check?
A background check is the verification of a person's or company's history before a decision: extending credit, leasing an asset or accepting risk. It typically combines identity, affiliations, court records, financial restrictions and behavioral history. In the U.S., when performed by a third party for employment, tenancy or credit purposes, it falls under the Fair Credit Reporting Act.
Why are court records often inaccurate in background checks?
Because they are assembled from hundreds of independent court systems with no shared taxonomy, and because the most common defects are structural rather than occasional: labels assigned at filing rather than resolution, missing dispositions, duplicate entries for one proceeding, and matching performed on name alone.
What does the FCRA require for reporting court records?
Section 607(b) requires reasonable procedures to assure maximum possible accuracy. The CFPB's January 2024 advisory opinion applied that standard to public records, addressing expunged and sealed records, missing dispositions, duplicates, and the presentation of multiple stages of a single proceeding. This article describes the data problem and is not legal advice. Compliance obligations depend on your role and use case.
Can a single court record disqualify an applicant?
It should not. An individual record is evidence, and the decision needs more than one data point. The meaningful signal is usually the pattern: repetition, the nature of the matter, which side of it the person was on, and how recent it is. A blanket rule against any court record is the easiest policy to write and the most expensive one to maintain.
Conclusion
The quality of a background check depends on whether anything downstream can read what its sources return, far more than on how many sources it queries.
As long as a risk decision rests on a code entered by a clerk in one of hundreds of unconnected court systems, the outcome carries that error forward. It declines people who should have been approved, misses the cases that mattered, and creates dispute exposure on both counts. The court has no obligation to get that label right for your purposes. You do have an obligation to get the report right.
See how it works in the verification stage.
Sources: National Center for State Courts, Court Statistics Project (2024 caseload data); The Pew Charitable Trusts, "State Courts Play a Key Role in American Life" (2024); CFPB, Advisory Opinion on Fair Credit Reporting: Background Screening, 89 Fed. Reg. 4171 (Jan. 23, 2024); WebRecon consumer litigation statistics (2025 to 2026).