Evidence Boundaries for Dispensary Document Retention

Research question

When can a retained dispensary document support a later review? The unit is one record version with a defined owner, source, and retention rule. The population is the record set requested for a review period. A file that exists but cannot be located, opened, attributed, or tied to the relevant event is not equivalent to readily retrievable evidence.

Virginia cannabis oversight, IRS recordkeeping guidance, NARA records-management principles, and NIST information-governance material support retention, authenticity, and retrievability. They do not establish one retention period for every cannabis document. The controlling rule depends on record type, jurisdiction, contract, tax status, and operational purpose.

Evidence and definitions

Store record type, source system, creation date, effective period, owner, access history where appropriate, retention trigger, destruction hold, and version relationship. A retention trigger might be creation, transaction completion, contract end, or an incident closure. Do not start every clock from upload date without checking the governing rule.

Measure current retrievability, metadata completeness, and version integrity separately. A record can be present but missing its effective date. A file can be retrievable but superseded. A current copy can be valid for today while the historical copy is needed to explain an earlier decision.

Interpretation and limitations

A high retrieval rate can coexist with weak metadata. A lower rate after a system migration may reflect indexing failure rather than lost records, or genuine loss. Test a defined sample and report selection method. Do not infer full repository quality from a convenient sample.

Transfer is bounded by the record’s authority and the jurisdiction. IRS guidance does not replace Virginia cannabis requirements. A retention policy is not proof that records were actually retained. Legal holds and privacy obligations can change the appropriate action; this research does not give legal advice.

Testing retrievability as an evidence property

Retention is not a single yes-or-no state. A document can be stored but not indexed, indexed but unreadable, readable but missing its effective date, or complete but inaccessible to the reviewer who has a legitimate need. Test these states separately. Define a requested record set, select a reproducible sample, and record whether each item was found, opened, attributed, interpreted, and linked to the relevant transaction or decision.

The retention trigger must be tied to the record’s meaning. A policy may run from approval, a transaction record from completion, a contract from termination, and an incident file from closure. Upload date is only a storage event. If multiple obligations apply, preserve the stricter or longer requirement according to the operator’s approved policy and current authority. Do not silently shorten a schedule because a system migration created duplicate-looking files.

Version control changes the question a record can answer. The current form may explain today’s process, while the historical version explains what a worker or reviewer saw at the time. Keep effective-from and effective-to dates, supersession reason, approver, and links between versions. A PDF export without its source metadata may be adequate for reading but weak for proving who approved a change. Conversely, a database row without a stable rendering may be difficult for an independent reviewer to interpret.

Measure missingness by field, not only by file. A repository can have a high file-retrieval rate and still lack owners, dates, or source identifiers. Report the numerator and denominator for each measure, plus unknown and not-applicable states. “Not applicable” should be a deliberate value with a reason, not a blank. For a migration, compare an inventory of expected records with the target inventory and separately report duplicates, orphaned records, failed conversions, and records awaiting legal-hold review.

Sampling should follow the risk and the claim. A random sample can estimate general retrievability; a targeted sample can test a known high-risk record type but cannot support a repository-wide percentage. Preserve the selection method and replacement rule. If a record cannot be opened because permissions are wrong, classify it as an access failure rather than missing. If it opens but the source cannot be established, classify it as provenance failure. These distinctions point to different corrective actions.

Record destruction is an event, not an absence. Retain the approved disposition authority, record category, trigger date, hold check, execution date, and operator or system identity. A destruction log can establish that a scheduled action occurred; it cannot prove that no unauthorized copy existed elsewhere. Legal holds, investigations, privacy requests, and contractual duties can interrupt ordinary schedules. Escalate those conditions to the authorized owner rather than improvising a deletion decision from a dashboard.

The transfer boundary is especially important for cannabis records. IRS guidance, NARA principles, NIST privacy material, and Virginia sources address different scopes and should not be merged into one invented schedule. A local result can show that a defined set was retrievable on a test date. It cannot certify compliance with every tax, regulatory, privacy, employment, or litigation requirement. The bounded conclusion is that retention research is strongest when the trigger, authority, version, access state, and test method remain visible.

A practical comparison boundary

Record the retrieval test date and repository state. A migration can make a record retrievable later without proving it was available when the review began. Preserve failed searches, permission errors, and subsequent recoveries as separate states. This gives the organization a defensible before-and-after measure without erasing the original evidence gap.

What a reviewer should request

Request the expected record inventory, retention rule and trigger, version map, retrieval results, permission failures, and hold or disposition status. A file list cannot establish that records are authoritative or interpretable. Preserve failed searches and recovered records as evidence of the retrieval process, not as noise to remove from the final measure.

Why retrieval tests need a known request

A retrieval test should name the record type, period, source, and reviewer’s permitted access. Searching for “all documents” is not reproducible and can conceal an incomplete request. Use a defined request set and record partial fulfillment. This makes a later improvement measurable while preserving the limits of what the original test actually asked the repository to provide.

Interpreting a retrieval change

An improved retrieval rate after migration may show indexing repair, not newly retained records. Compare the same request set, repository state, permissions, and version rules. Report records still unresolved and distinguish a recovered historical item from a newly created replacement. This preserves the original evidence gap while showing whether the corrective action worked.

A final evidence check

Before reporting retrievability, ask whether the requested population, retention trigger, version, permission state, and retrieval date are recorded. Ask whether failed searches and recovered records remain visible. If not, report a repository test limitation rather than treating a file count as proof that the applicable retention obligations were met.

Reading uncertainty correctly

An inaccessible record is not automatically a destroyed record, and a readable copy is not automatically the authoritative version. Preserve the failure state, permission context, source identifier, and version relationship. This lets the organization repair retrieval without rewriting the historical result or treating a replacement copy as evidence of original availability.

Bounded conclusion

Retention evidence is strongest when the trigger, rule, version, owner, and retrieval result are explicit. It can show whether a defined record set is reviewable. It cannot establish compliance with every applicable retention obligation from storage alone.

Sources

1. Virginia Cannabis Control Authority
2. IRS Recordkeeping
3. National Archives Records Management
4. NIST Privacy Framework