Body-worn camera video: retention, release, and storage costs

Reference article · 6 min read

Body-worn camera programs are usually approved on the strength of the hardware. The recurring cost, and most of the ongoing policy work, sits in what happens after the recording stops. Two decisions drive everything else: how long each kind of video is kept, and who can obtain a copy. They are not separable questions. Release rules generate redaction work, redaction generates additional copies, and both extend the life of files that a retention schedule would otherwise have purged.

A retention policy the evidence system cannot enforce is not really a policy. If the schedule assigns a short period to one category and a much longer one to another, but the system supports only a single global setting, or depends on an officer remembering to tag a recording correctly, the agency ends up keeping everything for the longest period and paying for it. Define the categories first, then confirm the system can apply them without manual effort and produce evidence that it did.

Sort video into categories before setting any retention period

Retention is meaningless applied to "video." It has to be applied to defined categories, each with its own period and its own release posture. Most agencies end up with some version of the following:

• Non-evidentiary. No arrest, citation, use of force, complaint, or collision. Usually the largest share of recorded hours by a wide margin.
• Evidentiary, routine. Tied to a citation, misdemeanor, or ordinary report where the case will close relatively quickly.
• Evidentiary, serious or pending. Felony cases, cases in active prosecution, and anything under appeal.
• Critical incident. Use of force, officer-involved shooting, in-custody death, pursuit and collision.
• Complaint and administrative. Anything connected to an internal affairs matter or a civilian complaint, including video the complainant has not yet identified.
• Accidental, test, and training. Kept briefly or reclassified, but worth a category so it does not silently inherit an evidentiary period.

Decide who assigns the category and when. Assignment at or near upload, with supervisory review on the serious categories, is far cheaper than reclassification months later.

Write the schedule against your actual legal obligations

Retention periods come from several directions at once, and they stack rather than substitute. In most states the controlling instrument is the approved records retention schedule, often administered by the state archives or library, and departures from it require approval. Many states have also enacted body-worn camera statutes that set minimum periods for particular categories and rules about disclosure. On top of that sit discovery obligations, the limitations periods for civil claims including federal civil rights actions, and protections for juveniles, victims, and medical information. Requirements change, and they vary considerably between states, so confirm the current text with your records officer, your state archives, and your legal counsel rather than relying on a summary.

Two practical points. First, write the schedule as a minimum and a maximum. Keeping video long past its retention period is its own liability and its own cost. Second, make deletion an affirmative, logged, scheduled action rather than something that happens when someone gets around to it.

Holds are where forecasts break

Every retention schedule is overridden by holds: pending prosecution, appeal, internal investigation, tort claim, litigation hold, and in many states an open public records request. Holds are the main reason actual storage consumption exceeds the projection built from retention periods alone.

Require that holds be applied at the individual recording level, survive recategorization, be releasable only with a logged reason and identified user, and be reportable. An agency should be able to answer, at any time, how many hours and how many terabytes are currently under hold and why. Without that report, the first year of unexpected invoices has no explanation.

Release drives the redaction workload, and the workload is measured in minutes of video

Requests arrive from prosecutors, defense counsel, the person recorded, civil attorneys, news organizations, and the public. Redaction obligations typically cover faces of uninvolved parties and minors, victim identity, medical information, addresses and phone numbers spoken aloud, and law enforcement data that cannot be disclosed.

The workload scales with minutes of video, not with the number of requests. One multi-officer incident can produce many hours across several cameras, all of which may require frame-level review. Automated detection reduces the work but does not eliminate human quality assurance, because a single missed face is a disclosure problem. Budget for staff time and for the derivative copies redaction creates, since the redacted version, the export sent to the prosecutor, and the original all occupy storage and all need their own retention treatment.

Model the storage bill before changing the policy

Steady-state storage for a category is approximately the daily hours uploaded, multiplied by the gigabytes per hour your cameras actually produce at your configured resolution, multiplied by the retention period. Measure the gigabytes per hour from your own uploads instead of using a datasheet figure.

Two consequences follow. Extending retention on a category multiplies that category's stored volume roughly in proportion, and because non-evidentiary video is usually the bulk of recorded hours, a modest change there moves the annual bill far more than a large change to critical incident retention. Also model the cost drivers that sit outside raw storage: replicated copies, retrieval or export charges, transcoding and analytics priced per hour, and per-user licensing. Run at least three scenarios before adopting a change.

What to ask vendors and put in the contract

• Consumption reporting. Storage and hours broken out by category, by hold status, and by trend, available to the agency without a support ticket.
• Enforceable schedules. Per-category retention applied automatically, with an audit record showing what was deleted, when, and under what rule.
• Deletion that is real. Ask in writing how purges propagate to replicas and backups, on what timeline, and whether a certificate of destruction is available.
• Exit terms. A defined price, format, and timeline for full export of video, metadata, audit logs, and redaction records. Settle this before signing, not at renewal.
• Price mechanics. What happens when camera count grows, when retention changes, and what any "unlimited storage" term means at renewal and after termination.
• Redaction licensing. Per seat, per hour, or per minute, and whether non-sworn staff can be licensed to do the work.
• Security posture. Immutable audit logging of view, share, download, and delete, alignment with criminal justice information security requirements where applicable, and any cloud authorization your state requires.

HQT works with state and local agencies on evidence and public safety systems, including retention design, integration, and the reporting that makes a retention policy auditable.

Frequently asked questions