The short version
Decide how long you need to keep footage, then size the storage to that number. Doing it the other way round, buying a recorder and accepting whatever retention it delivers, is the most common reason a business finds the footage it needs has already been overwritten.
For most businesses, 30 days is the right retention period. Sites where losses surface late, or where an insurer, licence or industry rule asks for more, keep 60 to 90 days. Storage then follows from four numbers: how many cameras, their average bitrate, how many hours a day they record, and how many days you keep.
| Question | Short answer |
|---|---|
| Typical retention for a business | 30 days |
| When to keep longer | Late-discovered losses, insurer or licence conditions: 60 to 90 days |
| Minimum for a Victorian liquor licence with a CCTV condition | 4 weeks |
| Storage formula | GB per camera per day = Mbps × 86,400 ÷ 8 ÷ 1,000 |
| One camera at 4 Mbps, 24 hours | 43.2 GB a day, 1.3 TB for 30 days |
| 16 cameras at 4 Mbps, 30 days | 20.7 TB of footage, before headroom and RAID |
| What UniFi Protect does when full | Deletes the oldest footage to make room (automatic retention) |
How long should business CCTV footage be kept?
Thirty days covers most incidents most businesses need to review. A break-in, a customer complaint, a damaged vehicle in the car park or a dispute at the counter is usually known about within days, and 30 days leaves time to find and export the footage.
Four things push the period longer.
How long it takes to discover an incident. This is the deciding factor more often than anything else. A retailer that finds stock missing at a monthly stocktake needs footage from up to five weeks earlier. A warehouse that discovers a short delivery when the customer complains a fortnight later needs the dock footage from before that. Work out the longest realistic gap between an incident and someone noticing it, then add the time it takes to review footage. If that is more than about three weeks, keep 60 days or more.
Insurance conditions. Some policies, particularly for high-value stock, cash or vehicles, set expectations for cameras and recording. Read the policy, or ask the broker, before choosing the retention period.
Licence and industry conditions. In Victoria, a liquor licence can carry a condition to install and maintain CCTV. Where it does, Liquor Control Victoria says recordings must be kept for at least 4 weeks and made available immediately to police or liquor inspectors who ask to see or remove them. Other regulated sites, such as those handling scheduled drugs or firearms, should check their own permit or licence conditions.
Privacy. Keeping footage longer is not automatically safer. Footage of identifiable people is personal information, and Australian Privacy Principle 11.2 requires a business covered by the Privacy Act to take reasonable steps to destroy or de-identify personal information once it no longer needs it for a permitted purpose. Pick a period you can justify, set the recorder to enforce it, and write it down. Business CCTV and the law in Victoria covers the wider obligations.
| Type of site | Usual retention | Why |
|---|---|---|
| Office | 30 days | Incidents are usually noticed within days |
| Clinic or allied health | 30 days | Short discovery time; sensitive footage argues against keeping more |
| Retail shop | 30 to 60 days | Stock loss may only show at stocktake |
| Warehouse or distribution | 60 to 90 days | Delivery disputes and shrinkage surface weeks later |
| Licensed venue with a CCTV condition | At least 4 weeks | Licence condition |
What decides how much storage CCTV needs?
Storage is the average bitrate of each camera, multiplied by the time it records, multiplied by the number of cameras and days. Everything else, including resolution, codec, frame rate and recording mode, matters because it changes one of those numbers.
| Factor | Effect on storage |
|---|---|
| Number of cameras | Storage rises in direct proportion |
| Resolution | More pixels means a higher bitrate for the same picture quality |
| Codec (H.264 or H.265) | H.265 compresses better than H.264; the saving depends on the scene |
| Frame rate | More frames per second means a higher bitrate |
| Scene activity | Movement, foliage, rain and traffic raise the bitrate; a still corridor compresses well |
| Night and infrared | Low-light noise can raise the bitrate at night |
| Recording mode | Continuous records 24 hours; detection-only records only events |
| Retention period | Storage rises in direct proportion to days kept |
Bitrate is the number that matters. It is the amount of data the camera sends per second, usually quoted in megabits per second (Mbps). Two cameras with the same resolution can run at very different bitrates depending on the codec, the frame rate, the quality setting and what they are looking at. Plan on the average bitrate each camera actually runs at over a busy day, not the figure on a data sheet.
H.265 versus H.264. H.265 (also called HEVC) was designed to give the same picture quality as H.264 at about half the bitrate. On CCTV the real saving is usually smaller than that and depends on the scene. In UniFi Protect, Ubiquiti lists its Enhanced Encoding setting as one way to get better compression and longer retention.
Frame rate. Business cameras rarely need 30 frames per second. 15 frames per second is enough to follow movement for most security purposes, and lowering the frame rate lowers the bitrate. For licensed venues with a CCTV condition, Liquor Control Victoria’s guidance asks for 8 frames per second or more.
Continuous or detection-only. UniFi Protect offers three recording modes per camera: Always, Detection Only and Never. Detection Only, with a few seconds of pre-event and post-event buffer, can cut storage sharply on quiet cameras. Keep continuous recording on cameras where you can’t afford to miss anything, such as entrances, tills and the cash office, because detection can miss a slow or partly hidden event.
The storage formula
Every figure in this guide uses one formula, so you can check it or rerun it with your own numbers:
GB per camera per day = average bitrate (Mbps) × 86,400 seconds ÷ 8 bits per byte ÷ 1,000 megabytes per gigabyte
That works out to 10.8 GB a day for every 1 Mbps of average bitrate, recording 24 hours. Multiply by the number of cameras and the days kept, then divide by 1,000 for terabytes. The table uses decimal units (1 TB = 1,000 GB), which is how drive capacities are labelled.
| Average bitrate per camera | GB per camera per day | TB per camera for 30 days |
|---|---|---|
| 1 Mbps | 10.8 | 0.32 |
| 2 Mbps | 21.6 | 0.65 |
| 4 Mbps | 43.2 | 1.30 |
| 6 Mbps | 64.8 | 1.94 |
| 8 Mbps | 86.4 | 2.59 |
| 12 Mbps | 129.6 | 3.89 |
| 16 Mbps | 172.8 | 5.18 |
CCTV storage table: 4 to 32 cameras at 30, 60 and 90 days
Assumptions: every camera averages 4 Mbps and records continuously, 24 hours a day. That is 43.2 GB per camera per day. The figures are footage only, before headroom and before the capacity RAID uses. They are a planning example, not a measurement of any particular camera; if your cameras average 8 Mbps, double every figure, and at 2 Mbps halve them.
| Cameras | GB per day (all cameras) | 30 days | 60 days | 90 days |
|---|---|---|---|---|
| 4 | 172.8 | 5.2 TB | 10.4 TB | 15.6 TB |
| 8 | 345.6 | 10.4 TB | 20.7 TB | 31.1 TB |
| 16 | 691.2 | 20.7 TB | 41.5 TB | 62.2 TB |
| 32 | 1,382.4 | 41.5 TB | 82.9 TB | 124.4 TB |
Checking one row: 16 cameras × 43.2 GB = 691.2 GB a day; × 30 days = 20,736 GB, or 20.7 TB.
Detection-only recording changes the hours, not the formula. A camera at 4 Mbps that records 6 hours of events a day writes 43.2 × 6 ÷ 24 = 10.8 GB a day. Most real systems mix the two: continuous on the entrances and tills, detection-only on the yard, corridors and store rooms.
How UniFi Protect records and stores footage
UniFi Protect records to hard drives or SSDs inside the UniFi console or network video recorder (NVR) on site, not to a cloud subscription. Retention is therefore limited by the drives you install.
When the drives are full. With automatic retention, Protect fills the drives and then deletes the oldest recordings to make room. Recording doesn’t stop, but the number of days you can look back is whatever fits. Protect also lets you set retention per camera, so a less important camera can keep fewer days and leave more space for the entrances.
Enhanced Retention. From Protect 5.1, Enhanced Retention keeps recent footage at full quality for a period you choose, such as 30 days, then downgrades older footage to low quality before it is deleted at the end of the final retention period. Downgraded footage can’t be restored to full quality. It suits sites that need a long look-back but only need detail for recent events.
Keeping footage beyond the retention period. Ubiquiti lists three ways: flag a clip to stop it being downgraded, export or archive it to external storage, or keep it in Case Manager. Do one of these as soon as an incident is known about.
Drive redundancy. Multi-bay UniFi NVRs combine drives in RAID so that a failed drive doesn’t lose footage. Ubiquiti’s Basic Protection uses RAID 1 for two drives and RAID 5 for three or more, and protects against one drive failing. Its larger enterprise NVR can use RAID 6 with nine or more drives, which survives two failures. Better Performance uses RAID 10 on an even number of drives. All of them use capacity:
| Ubiquiti protection option | RAID level | Survives | Usable share of raw capacity |
|---|---|---|---|
| Basic Protection | RAID 1 (2 drives) or RAID 5 (3 or more) | 1 drive failure | 67% to 75% with RAID 5 |
| Advanced Protection | RAID 6 (enterprise NVR, 9 or more drives) | 2 drive failures | 78% to 87% |
| Better Performance | RAID 10 (even number, at least 4) | Up to half the drives, one per mirrored pair | 50% |
Ubiquiti also notes that every drive in a RAID group must be the same size; a larger drive added to an array of smaller ones only contributes the smaller capacity until all of them are replaced. Single-bay consoles have no redundancy, so a drive failure loses the footage on it.
Recording drives work continuously and wear out. Ubiquiti recommends CMR drives at 7200 RPM for multi-bay consoles running RAID 5, 6 or 10. Plan for drive replacement over the life of the system.
How much headroom to allow
Size for at least 20% more storage than the table gives, then add the RAID overhead on top. Three things eat into the margin:
- Bitrates rise. Busier trading, a new tree in the frame, rain at night or a camera upgrade all increase the average.
- Cameras get added. Most systems grow. Storage sized for today’s count runs short with the first extra camera.
- Retained clips take space. Flagged clips and anything kept for a dispute sit outside the rolling retention.
Worked example: 16 cameras at 4 Mbps for 30 days is 20.7 TB of footage. With 20% headroom that is about 24.9 TB. In RAID 5 with four equal drives, 75% of raw capacity is usable, so the drives need about 33 TB raw between them. After installation, check how many days the system is actually holding against the target, and check it again a month later.
Exporting footage for police and insurers
Footage is only useful if it is still there and can be shown to be genuine.
- Act fast. Export the relevant clips as soon as the incident is known, before rolling retention deletes them.
- Check the clock. Confirm the recorder’s time is correct and note any offset. Time synchronisation should be set up at installation.
- Export the original. Keep the exported file unedited. Make copies for police, the insurer or a lawyer, rather than giving anyone access to the recorder.
- Write it down. Record the camera name, date, time range, who exported it, and who received copies and when.
- Keep it only as long as needed. Once the matter is closed, delete exported copies in line with the retention policy.
Whether you may share footage at all, and with whom, is a legal question as well as a technical one. Business CCTV and the law in Victoria covers sharing footage with police.
A sizing checklist before you ask for a quote
- Decide the retention period, and why, with the insurer and any licence or industry conditions in mind.
- List every camera, current and planned, and mark which record continuously and which on detection.
- Use 4 Mbps per camera as a starting assumption, then replace it with the real bitrate once cameras are chosen.
- Apply the formula, add at least 20% headroom, and add the RAID overhead.
- Choose a recorder with enough bays to reach that capacity with redundancy, and room to grow.
- Decide who can view and export footage, and how exports are logged.
The number of cameras comes first. How many security cameras a business needs covers the count and positions, and UniFi Protect versus a traditional NVR covers the recorder choice. When you’re ready, security systems and cameras starts with a site walkthrough and a quote sized to the retention you need.
