What an adjuster sees that you don't
Adjusters and forensic-video examiners read footage as evidence under specific standards: pixels-on-target for identification, retention covering the discovery window, chain of custody from recorder to delivery. The five mistakes below come up over and over in declined claims and adverse premises-liability outcomes. Each has a concrete fix.
1. Coverage gaps where the incident actually happened
Cameras at the front door, nothing on the side hallway, nothing in the back-of-house. The incident happens in the spot the camera doesn't see. Adjusters ask for the clip and the answer is 'we don't have that angle.' The fix is a coverage walk that maps every space against its threats: entries, transition zones, registers, stockrooms, slip zones, parking. Most properties find three to seven coverage gaps; not all need new cameras, sometimes a re-aim solves it.
2. Resolution below the readable floor
720p covering a 200-foot run produces a face six pixels wide. Adjusters and forensic-video examiners use pixels-on-target: 80 pixels across the face for identification, 200 across a plate. Cameras below the threshold produce footage that supports a story but doesn't identify the person. 1080p is the practical floor for general use; 4K at chokepoints handles identification.
3. Retention that drops the clip before discovery
A slip-and-fall reported on day 25 needs the footage to still exist. A theft pattern noticed at quarterly inventory needs 60 days of history. Systems set to 7 or 14 days lose the evidence before the loss is reported. PCI-DSS mandates 90 days for card-data environments. The safe commercial target is 90 days; cloud-managed VMS handles it without on-prem storage management.
4. No chain of custody on the export
The footage exists, but the export shows up on a USB drive with no metadata, no access log, no hash. Defense counsel argues the file could have been altered between recorder and courtroom. The judge agrees. The fix is digital watermarking or hashing on the original, an access log showing who exported what and when, and a written custody timeline. Modern cloud VMS handles this automatically; legacy NVRs require a documented manual workflow.
5. Recorder failures with no redundancy
The on-prem recorder fails or gets destroyed in the same incident the footage was supposed to capture. Power blip, lightning, fire, theft of the recorder itself. No cloud copy, no hybrid backup, no footage. Cloud-managed VMS writes a parallel cloud copy that survives any local event. Hybrid configurations on Genetec or Milestone keep the on-prem recorder primary with cloud as backup. The premium runs $20 to $60 per camera per month and pays for itself the first time the recorder dies.
The retention floor by industry
PCI-DSS: 90 days for any merchant or service provider handling card data. HIPAA-touching healthcare: 30 to 90 days, workflow-dependent. FERPA-touching education: 30 to 90 days. NDAA-touching federal-grant-funded sites: matched to the grant retention requirement, often 90 days. General commercial: 30-day floor, 90-day target for properties with claim history.
We work multi-vendor across Verkada, Avigilon Alta, Genetec, Axis, Hanwha, Eagle Eye Networks, Milestone, and Rhombus, plus camera-agnostic analytics on Intenseye and Dragonfruit AI.