North Dakota Monthly Production Revision Lineage Study 2026: Research question and study design
Monthly production data can change after first publication. The defensible research object is therefore not “the volume,” but a four-part observation: well, product, production month, and publication vintage. This study uses a vintage-panel design to compare what an official source displayed at different access times.
The method does not presume that every difference is a correction or that every stable value is final. It records revisions, additions, removals, and presentation changes while reserving accounting, royalty, operational, and regulatory explanations for evidence that actually addresses them.
Freeze each vintage
Preserve the downloadable file or reproducible query result when permitted, along with retrieval time, route, filters, file size, and cryptographic hash. Record the publisher’s labels and any notes accompanying the release. A later download with the same filename is a new observation if its bytes or contents differ.
Do not overwrite the earlier extract. Without both vintages there is no evidence of revision—only a current value and a recollection. If retention is restricted, save sufficient metadata and a lawful locator to reproduce the observation.
Construct the panel key
Key rows with the jurisdictional well identifier, explicitly named product, production month, and vintage. Keep product categories separate; oil, gas, and water figures are not interchangeable. Retain units and status flags. A normalized well key may assist joining, but the original identifier remains beside it.
Duplicate keys require investigation rather than automatic summing. They may represent multiple report lines, differing products, corrections, or a malformed extract. Aggregation is an analytical decision that needs a stated rule and reviewer.
Classify deltas before interpreting them
Comparison produces several useful classes: unchanged value, changed value, newly appearing row, disappearing row, and representation change. Representation covers blank-to-zero, text-to-number, unit-label, or code changes that may not represent a physical volume difference.
Calculate absolute and percentage deltas only when both values are comparable and the denominator is meaningful. Preserve the raw pair. A delta demonstrates that two official observations differ; it does not identify late reporting, allocation, audit action, correction, or system maintenance as the cause.
Distinguish zero from absence
A reported zero is an observed value. A blank may be an empty field. A missing row means the panel key was not present in that vintage. A suppressed or coded value carries the publisher’s stated convention. Converting all four states to zero manufactures evidence.
Negative searches follow the same rule. Record the query, scope, and return condition. “No row returned for this key in this extract” is precise. “The well produced nothing” exceeds what the search can establish.
Example of a two-vintage panel
Assume the first extract shows oil volume for a well and month, leaves gas blank, and contains no water row. The second shows a different oil volume, gas as zero, and a new water row. The panel records one numerical revision, one representation transition, and one addition.
It does not net those changes into a single adjustment or assert a cause. The review packet includes both file hashes, source rows, extraction logic, calculated deltas, and questions for accounting or regulatory staff. If royalty consequences matter, separate ownership and payment evidence must be obtained.
Release ledger for reproducibility
Maintain a release ledger beside the panel. List every acquired vintage, locator, retrieval timestamp, byte hash, parser or query version, row count, duplicate count, and import warning. This distinguishes a publisher revision from a change in extraction logic.
When comparison code changes, rerun both vintages under one version before interpreting new deltas. Otherwise a formatting fix can look like a production revision. Save the classification rules for blank, zero, missing, and coded states and keep known examples as regression checks.
Controls, exclusions, and result
Reperformance should begin from the preserved vintages and code or formulas used for the join. A checker confirms record counts, duplicate handling, units, key construction, and a sample of changed and unchanged rows. Large deltas may receive priority, but materiality thresholds should be supplied by the decision owner.
This study assessed public source routes on October 6, 2026. It did not determine measurement accuracy, allocation, sales, taxes, ownership, proceeds, deductions, or payment entitlement. Federal statistics may help with broad context but are not substitutes for North Dakota well-month records. The core result is that vintage panels preserve change honestly and stop unexplained differences from becoming invented explanations.
Field protocol
Treat well API as part of the panel key or measured observation, not free-floating metadata. Compare it only across compatible vintages, preserve its raw state, and document the rule used for every calculated difference. The audit row should also identify the extraction version, any import warning, and the reviewer who accepted or rejected the delta.
Treat product code as part of the panel key or measured observation, not free-floating metadata. Compare it only across compatible vintages, preserve its raw state, and document the rule used for every calculated difference. The audit row should also identify the extraction version, any import warning, and the reviewer who accepted or rejected the delta.
Treat production month as part of the panel key or measured observation, not free-floating metadata. Compare it only across compatible vintages, preserve its raw state, and document the rule used for every calculated difference. The audit row should also identify the extraction version, any import warning, and the reviewer who accepted or rejected the delta.
Treat publication vintage as part of the panel key or measured observation, not free-floating metadata. Compare it only across compatible vintages, preserve its raw state, and document the rule used for every calculated difference. The audit row should also identify the extraction version, any import warning, and the reviewer who accepted or rejected the delta.
Treat reported unit as part of the panel key or measured observation, not free-floating metadata. Compare it only across compatible vintages, preserve its raw state, and document the rule used for every calculated difference. The audit row should also identify the extraction version, any import warning, and the reviewer who accepted or rejected the delta.
Treat raw volume as part of the panel key or measured observation, not free-floating metadata. Compare it only across compatible vintages, preserve its raw state, and document the rule used for every calculated difference. The audit row should also identify the extraction version, any import warning, and the reviewer who accepted or rejected the delta.
Treat status code as part of the panel key or measured observation, not free-floating metadata. Compare it only across compatible vintages, preserve its raw state, and document the rule used for every calculated difference. The audit row should also identify the extraction version, any import warning, and the reviewer who accepted or rejected the delta.
Treat file hash as part of the panel key or measured observation, not free-floating metadata. Compare it only across compatible vintages, preserve its raw state, and document the rule used for every calculated difference. The audit row should also identify the extraction version, any import warning, and the reviewer who accepted or rejected the delta.
Treat row presence as part of the panel key or measured observation, not free-floating metadata. Compare it only across compatible vintages, preserve its raw state, and document the rule used for every calculated difference. The audit row should also identify the extraction version, any import warning, and the reviewer who accepted or rejected the delta.
Treat absolute delta as part of the panel key or measured observation, not free-floating metadata. Compare it only across compatible vintages, preserve its raw state, and document the rule used for every calculated difference. The audit row should also identify the extraction version, any import warning, and the reviewer who accepted or rejected the delta.
Treat percentage delta as part of the panel key or measured observation, not free-floating metadata. Compare it only across compatible vintages, preserve its raw state, and document the rule used for every calculated difference. The audit row should also identify the extraction version, any import warning, and the reviewer who accepted or rejected the delta.
Treat review disposition as part of the panel key or measured observation, not free-floating metadata. Compare it only across compatible vintages, preserve its raw state, and document the rule used for every calculated difference. The audit row should also identify the extraction version, any import warning, and the reviewer who accepted or rejected the delta.
Sources
Frequently asked questions
Is the newest value always correct?
The newest value is the latest observed vintage; correctness or finality requires additional evidence.
Should blank values become zero?
No. Preserve blank, zero, missing row, and publisher-coded states separately.
Can a revision prove underpayment?
No. Payment conclusions require ownership, pricing, deductions, timing, and accounting evidence beyond production volume.
If your team needs a review-ready north dakota monthly production revision lineage study packet, Book a free consultation.
