A municipal bus-surveillance tender asked for two retention numbers: at least 240 hours of video on every vehicle, and at least 13 months in the central archive.
Those numbers came from the buyer, writing a procurement document. That is how fleet video retention is often set: not by a regulator handing down a figure, but by someone deciding — or, far more often, by nobody deciding, and the recorder's storage settling it by default.
This post is about making that decision on purpose. It makes no claim about what any law requires; what each regulatory regime demands of video recording is covered separately. The question here is practical: who and what should set the number, and how to write it down.
The default: the recorder decides
Every vehicle recorder overwrites its oldest footage when it fills up. That means every fleet already has a retention policy, whether or not anyone wrote it: the recorder's buffer horizon.
For one current two-camera dashcam, that horizon is about 36 hours on a 256 GB card at Normal quality. A heavier mobile NVR with a 2 TB drive holds roughly 13 days at a similar load. How the buffer is sized, and what it still loses, is its own subject — but the point for retention is blunt. If nobody sets a policy, the fleet's retention period is "until the card fills", which can be as little as a day or two.
Who actually sets the number
Four forces decide fleet video retention in practice. Only one of them is usually written down.
Force | What it pushes toward | Where it comes from |
|---|---|---|
The buyer or contract | A specific minimum, on the vehicle and centrally | Tenders, customer contracts, internal policy — e.g. the 240 hours / 13 months above |
The claims and complaints window | Longer — footage must outlive the gap between an incident and someone asking for it | Insurers and customers use the footage; none of the insurer guidance we found states a period |
Storage cost | Shorter, and more selective | Grows with days kept × volume kept |
Privacy law | Shorter — keep only as long as the purpose needs | India's DPDP Act, for one, treats footage of identifiable people as personal data |
The second row is the most useful finding in this post, and it is an absence. Insurers clearly value fleet footage — Aviva's fleet risk guidance notes it "can also be used to identify who may be at fault in a collision" — but we could find no insurer guidance that tells a fleet how long to keep it. The insurer asks for the clip when there is a claim. Whether the clip still exists is the fleet's problem.
So the claims window has to be estimated by the fleet itself, from its own history: how long, in practice, between an incident and the first request for footage? That number, not a guess, is the floor for anything that might become evidence.
Why "keep everything" is the expensive answer
The instinctive response to that uncertainty is to keep everything, for as long as possible. The arithmetic says otherwise.
A two-camera recorder at 14 Mbps writes about 63 GB in a ten-hour shift; across 200 vehicles that is 12.6 TB a day. Compare two policies for the central archive, each kept for 13 months:
Policy (200 vehicles, 13 months) | Data held at steady state | At S3 Standard list price |
|---|---|---|
Full continuous recording | ≈ 4,977 TB | ≈ $114,000 per month |
Event clips only (20 × 90 s per vehicle per month) | ≈ 8.3 TB | ≈ $190 per month |
Priced at S3 Standard's $0.023 per GB-month for scale only; archive tiers are far cheaper, and the 24/7 bill your retention policy controls works through tiering properly. The event count is an illustrative assumption. The ratio — roughly 600 to one — is the point, and it holds at any storage price.
Most of a fleet's recording never needs to reach the archive at all. Which footage leaves the vehicle, and by which route, is the offload decision; retention is what happens to it once it arrives.
Privacy pushes the other way
Cost pushes toward keeping less. So does data-protection law, for a different reason: footage of drivers, passengers and bystanders is information about identifiable people.
India's Digital Personal Data Protection Act states the principle directly. Under section 8(7), a data fiduciary shall, "unless retention is necessary for compliance with any law for the time being in force", erase personal data "as soon as it is reasonable to assume that the specified purpose is no longer being served".
We quote it as a direction of travel, not as advice. Its practical effect on a retention policy is to make "we might need it someday" a weak reason on its own. Each tier of footage should be kept for a stated purpose, for as long as that purpose lasts — which is also, conveniently, what keeps the storage bill down.
A tiered retention policy
Put the four forces together and a defensible policy has three tiers and one override.
- Tier 1 — on the vehicle. Continuous recording, kept for the recorder's horizon. Size that horizon to outlast your longest realistic gap before footage can be requested or offloaded.
- Tier 2 — event footage, centrally. Clips around triggers and anything requested. Kept for the longer of your contractual minimum and your measured claims window.
- Tier 3 — bulk footage, centrally, if at all. Short, or zero. Keep only what a contract or a stated purpose requires.
- The override — preservation. Once an incident is flagged, its footage leaves the deletion schedule until the matter closes. That footage also needs timestamps, retention and export that hold up later.
Write each tier down with its number, its purpose and its owner. A retention policy is only real when someone could be asked why a clip was deleted, and answer.
The Bottom Line
Fleet video retention is set by contracts, claims windows, storage cost and privacy law — and when nobody sets it, by the recorder's card. Insurers value footage, but the insurer guidance we found gives no retention period, so the claims window must come from the fleet's own history.
Keep continuous footage on the vehicle, event footage centrally for as long as its purpose lasts, bulk footage briefly or not at all, and preserve anything flagged. The difference between that and "keep everything" is roughly 600 to one.
What's Next
Retained footage is only as useful as its custody: a record button isn't a chain of custody covers what has to be true for a clip to hold up. And for the full path that footage travelled before it reached the archive, see the architecture between vehicle and command centre.
Frequently Asked Questions
How long should a fleet keep dashcam footage?
There is no single number. Keep continuous footage on the vehicle for the recorder's horizon, keep event and requested footage centrally for the longer of your contractual minimum and your measured claims window, and keep bulk footage briefly or not at all. Preserve anything tied to a flagged incident until it closes.
Who decides fleet video retention?
In practice, four forces: the buyer or customer contract, the time between incidents and requests for footage, storage cost, and data-protection law. When none of them is written into a policy, the recorder's storage decides by overwriting the oldest footage.
Do insurers require fleets to keep video for a set period?
None of the insurer guidance we reviewed states a retention period. Insurers use fleet footage to establish fault, and ask for it when there is a claim; whether it still exists is the fleet's responsibility. Set retention from your own claims history.
How much does it cost to retain fleet video?
It depends almost entirely on what you keep. For 200 vehicles over 13 months, retaining the full continuous recording is roughly 5,000 TB, while keeping only event clips is roughly 8 TB — about 600 times less, at any storage price. Samvyo, based on SFU architecture, keeps recording processing and storage on infrastructure you control, so tiering and retention periods follow your policy rather than a provider's default.
Does privacy law affect how long fleet video can be kept?
It pushes toward keeping less. India's DPDP Act, for example, requires personal data to be erased once its specified purpose is no longer served, unless another law requires retention. This is not legal advice; state a purpose and a period for each tier of footage.
What is a video retention tier?
A category of footage with its own storage location, retention period and purpose — for example, continuous footage on the vehicle, event clips in a central archive, and flagged incidents under preservation. Tiers let a fleet keep what matters for as long as it matters without paying to store everything. Platforms such as Samvyo that keep recording on infrastructure you choose let you set those periods per tier yourself.
