Research 2026

Detention Evidence Completeness: A Sampling Study Design

A sampling design for testing whether detention claims contain the timestamps and source evidence a billing team needs to review them.

Detention Evidence Completeness: A Sampling Study Design editorial illustration

Published October 22, 2026.

Research question: Which evidence fields are missing most often in detention claims, and how can a carrier or brokerage measure completeness without inventing waiting time?

Method and evidence scope

This paper defines a sampling design, not an observed benchmark. The unit is a closed detention claim. Required fields are arrival, gate check-in, dock assignment, work start, work stop, release, and a source for each timestamp. The design uses three strata: claims under two hours, claims between two and four hours, and claims above four hours. A pilot of 20 to 30 claims can reveal whether a field is broadly missing; a precise rate requires a larger designed sample. The sources reviewed describe freight operations and driver hours, but they do not provide a detention completeness standard.

Table of contents

Why completeness matters more than claim size

A large claim with a clean timeline can be reviewed quickly. A small claim with a missing release time can consume the same review effort and still fail. Completeness is a property of the evidence set, not of the dollar amount. Measuring it separately prevents a team from concluding that detention review is slow when the actual problem is an incomplete packet.

The seven timestamp fields

Arrival, gate check-in, dock assignment, work start, work stop, and release are the core sequence. The seventh field is the source for each timestamp: a gate log, a facility portal, a driver message, a photo, or a signed document. A timestamp without a source is an assertion. A source without a timestamp is an unattached document. Both should count as defects.

Defect rules before review

Write the rules before sampling. Missing means the field is absent or marked unknown. Invalid means the timestamp precedes the previous event, falls outside the delivery date, or conflicts with another source. Unverifiable means the source cannot be located. Decide how to treat a driver estimate and a facility rounding convention. Keep the rules stable for the whole sample.

Stratify by claim duration and facility

Stratify by claim duration because longer waits often have more handoffs and more gaps. Stratify by facility when the sample allows, because a single dock can produce a repeating pattern. Report the population count in each stratum before sampling. A sample dominated by short claims will understate the problem in long waits.

Sampling and second review

Use a random or consecutive sample within each stratum. Have a second reviewer check a subset and compare defect judgments. If disagreement is high, the defect rules are not specific enough. Record the clarification and apply it to the rest of the sample. Do not change the rules midway without marking the break.

Reporting and remediation

Report the claim count, defect rate per field, examples without personal data, the owner of the fix, and a review date. A defect pattern at the dock suggests a capture problem. A pattern by driver suggests a training or tool problem. A pattern by broker suggests a submission requirement that was never explained. Each has a different fix.

A minimum viable specification

Name the metric, the unit (closed claim), the seven timestamp fields, the source rule, the three duration strata, the exclusion for claims with no delivery record, the open-claim handling, and the review period. Store the specification with the report. When a broker changes its submission requirement, update the specification before comparing the next period. A fixed specification turns a one-time cleanup into a repeatable control.

Review cadence and ownership

Review the sample quarterly and after a change in brokers, facilities, or capture tools. The billing owner owns the claim decision; the dispatch owner owns timestamp capture; the assistant owns the queue and the sample preparation. Record the review date and the one process change selected for the next period. A report without an owner becomes a recurring complaint.

Limits

The design measures whether evidence is present and usable, not whether the claim is valid or the amount is correct. It depends on the claim file being the actual packet sent to the broker. If the packet is assembled after the claim is disputed, the measure describes the reconstruction, not the original evidence. It also does not establish that better evidence improves recovery rates; that requires a separate study.

Sources

  1. FMCSA: Hours of Service
  2. Bureau of Transportation Statistics: Freight
  3. OSHA: Trucking

Frequently Asked Questions

What counts as a complete detention claim?

The full timestamp sequence plus a locatable source for each event, with gaps marked unknown rather than estimated.

Should claims under two hours be included?

Yes, if they are part of the population. Stratify them so they do not hide defects in longer claims.

Can a driver statement substitute for a missing timestamp?

It can be recorded as a source, but it should not be presented as a facility timestamp. Label the evidence type.

How often should the sample run?

Quarterly for a stable process, and after a change in facilities, brokers, or capture tools.