Verkada Camera Monitoring as a Managed Service: Closing the 60-Second Detection Gap

After deploying HTS Sentinel across 6 sites and 50 cameras, a multi-site healthcare provider now detects every camera outage and tamper event in under 60 seconds, and ships a weekly proof-of-monitoring report to its insurance broker. Here is what a managed Verkada partner actually delivers.

The detection gap is real, and it is unowned

Two facts about enterprise video surveillance are not in dispute.

First, cameras fail silently. Industry observability data puts the median mean time to detect (MTTD) for a high-impact outage at 37 minutes, with 29% of organizations reporting MTTD of an hour or more [1]. A 2024 industry benchmark found that 44% of businesses take 30 or more minutes to detect a critical issue, and 60% take 30 or more minutes to resolve one [2]. For a security camera, "down" is not a degraded service tier; it is a silent gap in evidence nobody notices until the footage is needed.

Second, Verkada's native tooling covers about half the problem and none of the customer surface. The Device Stats dashboard (released October 9, 2025) reports online, offline, updating, and new status [3]. The native Tamper Alerts feature is accelerometer-based — it triggers on physical vibration or movement of the camera housing, not on what the camera is actually seeing [4]. A camera that has been defocused, taped over, or repositioned by one degree does not move its housing enough to trip an accelerometer. And Verkada's alerts live inside Command; they do not push to Slack, Microsoft Teams, PagerDuty, or a customer's general inbox.

The consequence: the buyer owns the part that matters most — routing the alert to a person and documenting that the alert was acknowledged. Most multi-site operators do not have a 24/7 security operations center to own that. A managed Verkada NOC does.

Why image-based detection outperforms the accelerometer

Sentinel runs six detection passes per camera every 30 seconds. The first five run on every poll; the sixth runs hourly.

The blocked, redirected, and defocused detectors look at the content of the frame, not the housing. Published research on computer-vision tamper detection shows 95.8%–97.3% detection rates on covered, moved, and defocused events at 20–30 fps, using edge-disappearance and Laplacian methods [5]. A camera that is moved one degree to the right will not move its housing enough to trip an accelerometer; it will change the pixel composition of the entire frame and trip a redirected alert within one polling cycle.

A 10-minute cooldown per (camera, kind) tuple prevents alert storms. A camera that reboots three times in five minutes generates one alert, not forty.

The 60-second budget and the customer surface

The detection pipeline is deterministic. For a single camera, the worst-case round trip from anomaly to customer alert is:

t=0s        Poll fires. Fetch status + image.
t=0–10s     Compare against rolling baseline.
t=10s       Detector fires (or does not fire).
t=10s       Cooldown check. Enqueue alert.
t=10–30s    Alert dispatcher routes to Slack / Teams / email / PagerDuty / SMS.
t=30s       Next poll.

The 99.95% pipeline-uptime SLA is met by a dual-region FastAPI deployment in AWS ECS Fargate (us-east-1 primary, us-west-2 backup) with multi-AZ Postgres and ElastiCache Redis. Webhook signature verification on the Verkada side uses HMAC-SHA256 over {timestamp}.{body} with a 5-minute timestamp tolerance. The 30-second poll is half the budget; the rest is detector evaluation, cooldown, and outbound routing.

The customer-facing surface is what a managed Verkada partner actually owns. Three artifacts.

The per-camera uptime dashboard. A grid view of every camera, organized by site, color-coded by health. Click a camera to see its last 7 days of uptime, anomaly events with timestamps, and a frame thumbnail from the moment the anomaly fired. Ninety-day retention is standard; 365 days on enterprise tier.

Alert routing configuration. Self-service, per-site, across Slack, Microsoft Teams, email, PagerDuty, and SMS. Each (site, kind) pair routes independently — the parking lot can be Slack-only during business hours and PagerDuty after hours, while the lobby stays email-only.

The weekly insurance-grade PDF. Auto-emailed every Monday at 07:00 local. One page per site: uptime percentage, anomaly counts, per-camera health score, MTTA and MTTR, tamper event log with frame thumbnails, and a Compliance section listing which insurance-discount criteria the deployment meets. This is the document a broker takes to a renewal meeting.

The economics: insurance, MTTD, and missed-incident exposure

For a 100-camera, 5-site operation, three lines of value are worth quantifying.

Insurance premium reduction. The widely repeated "5 to 20%" discount figure traces to security-vendor blogs rather than carrier filings, and the real rating factors are building, occupancy, loss history, and protection class [6]. The defensible range for documented active monitoring on a commercial property policy is 5–10% for a monitored burglar alarm, 5–15% for monitored intrusion plus 24-hour video, and 15–20% for a fully monitored suite with quarterly compliance documentation [7][8]. For a 5-site customer paying $250,000 a year in property and inland marine premiums, a 10% discount is $25,000 a year. Sentinel for 100 cameras at the Growth band ($8 per camera per month) is $9,600 a year. The insurance discount alone is 2.6× the service cost.

Mean time to detect and recover. New Relic's 2024 data puts median MTTD at 37 minutes and median MTTR at 51 minutes [1]. The reduction in blind-spot exposure per camera-out event is, conservatively, 35 minutes. Across 100 cameras at a conservative 4 camera-out events/camera/year, that is roughly 233 hours of avoided camera downtime per year.

Missed-incident avoidance. A single missed-incident liability event at a multi-site healthcare or education customer has historically settled in the $50,000 to $500,000 range, before counting reputational cost. Sentinel does not prevent the incident; it prevents the silent failure of the evidence chain when the incident is reviewed weeks later.

Worked example, 100 cameras, 5 sites, $250K annual property premium:

Insurance return on the service cost: 260% in year one, before counting exposure reduction.

What changed at a multi-site healthcare provider

The healthcare provider is a six-site, 50-camera specialty medical group with approximately 35 access-control readers, approximately 350 active badges, and roughly 120 NOC alerts per month flowing through the HTS Managed console. The customer's security director owned the relationship before Sentinel existed; HTS had been monitoring the Verkada deployment as part of an internal Verkada NOC.

Before Sentinel, the security director checked the Verkada Command dashboard for camera health on a best-effort basis during business hours. A camera offline on a Saturday night was detected on Monday morning — if at all. The KOP-04 parking lot had two real blind-spot events in 2025 — a hit-and-run and a vehicle break-in — and in both cases the relevant camera had been defocused for an unknown period before the incident. At the last renewal, the carrier asked whether the system was "actively monitored." The answer at the time was "no, in the formal sense."

Today, all 50 cameras are polled every 30 seconds. The detection-to-alert round trip on the parking-lot cameras is under 60 seconds — the alert lands in the security director's Slack at 11:47 PM on a Saturday within a minute of a defocus event. The parking-lot coverage was upgraded to a tamper-tolerant mount, and the 2025 hit-and-run is now documentable end-to-end. The relationship represents $6,646 in monthly recurring revenue to HTS, and Sentinel wraps the existing work in a co-branded product surface without changing the underlying economics.

Where Sentinel fits in the managed Verkada picture

A managed Verkada partner does not just watch cameras. Sentinel is one of five products on the HTS Managed platform and shares a single customer dashboard, alert-routing configuration, and ConnectWise agreement line with the rest. The full suite is the wedge; Sentinel is the entry point. BadgeFlow prints and ships badges. AccessOps handles HRIS-native identity. NetOps manages Meraki, Cambium, and Peplink networks. Comply packages the weekly PDF into a SOC 2 / HIPAA / PCI auditor portal. Sold together on one agreement, they replace four vendors and one internal spreadsheet.

The HTS-managed approach to Verkada camera monitoring: 60-second detection, image-based detectors, customer-facing alert routing, and a weekly insurance-grade PDF — at $4–$10 per camera per month.

Talk to us about a 50-camera pilot.

Sources

[1] New Relic, 2024 State of Observability, October 2024. https://newrelic.com/resources/report/observability-forecast/2024/state-of-observability/outages-downtime-cost

[2] Paessler AG, MTTD/MTTR Key Metrics for Effective Incident Response, 2024. https://blog.paessler.com/mttd-and-mttr-key-metrics-for-effective-incident-response

[3] Verkada, Device Stats Dashboard (Help Center), released October 9, 2025. https://help.verkada.com/verkada-cameras/analytics/camera-stats/device-stats-dashboard

[4] Verkada, Tamper Alerts (Help Center). https://help.verkada.com/verkada-cameras/analytics/create-camera-event-alerts/tamper-alerts

[5] Gil-Jiménez et al., Unified Camera Tamper Detection Based on Edge and Object Information, PMC4481966. https://pmc.ncbi.nlm.nih.gov/articles/PMC4481966/

[6] Surveillant.ai, Do Security Cameras Lower Business Insurance? The Real Answer. https://surveillant.ai/guides/do-security-cameras-lower-business-insurance

[7] Philibert Security, Commercial Alarm Insurance Rules. https://philibertsecurity.com/learn/commercial-insurance-requirements.php

[8] Twin City Security Denver, Insurance Carriers Are Raising the Bar on Property Security Standards, 2024. https://www.twincitysecuritydenver.com/insurance-carriers-are-raising-the-bar-on-property-security/

Harris Technology Services (HTS) · https://www.hts.pro · Support: support@hts.pro