Anonymized real case: a Mexican convenience store chain (14 stores, 800-1500 m² each, ~1,200 employees total) migrated from a legacy analog DVR system (2014) to a modern IP system with 200 cameras and centralized VMS. Here are the real numbers: costs, downtime, ROI, and the three things that went wrong.
( nominal figures — the chain requested not to publish by name, but the proportions and lessons learned are shareable )
Initial situation (Q3 2024)
- 14 stores, 1,200 m² average
- 200 analog cameras (mix of 4 brands, none NDAA, none HD)
- 14 independent DVRs (one per store), generic brand, no integration
- 1-month retention in DVR, then overwrite
- No central VMS, each store reviewed locally
- No analytics — only recording
- Shrink reported ANTAD 2024: 1.08% of sales (figure the chain replicated internally)
Annual operating cost of the legacy system:
- DVR maintenance: ~$8,000 MXN/store/year (labor + parts)
- Incident investigation: ~120 person-hours/month at corporate level, equivalent to ~$360,000 MXN/year
- Average time to find incident in footage: 45 minutes
- Loss from unresolved incidents: internal estimate ~$2.4M MXN/year (organized internal shrink incidents + POS errors without video backup)
Total annual operating cost legacy: ~$3.5M MXN (~USD $175,000).
Upgrade decision
Three drivers:
- Growing shrink. The chain detected that shrink in stores with legacy DVR exceeded the ANTAD average by 0.3 points. For 14 stores with annual aggregated sales ~$1,800M MXN, 0.3 points = $5.4M MXN/year.
- Investigation capacity. 45 minutes per incident is unsustainable when you have 8-10 incidents/week across 14 stores.
- Regulatory. The chain serves B2B corporate clients that began requiring NDAA compliance in their supply chain.
Proposed design (Q1 2025)
Technology stack:
- 200 4K IP cameras Hanwha XNV-9083R (vandal dome) + 30 2MP cameras in low-criticality zones (back office, warehouse). Total: 230 cameras.
- NVRs in each store: Hanwha XRN-1620 (16 channels, 2U) or XRN-2010 (32 channels, 2U).
- Central VMS: Genetec Security Center 5.11 (per-camera licenses + Omnicast).
- Network: managed PoE+ switches Ubiquiti USW-24-PoE in each store, dedicated 100 Mbps uplink to corporate via IPsec.
- Local storage: 30 days per store, RAID 5 with 4×10 TB.
- Centralized storage: zero — everything stays in store; corporate accesses by streaming from NVR.
Cost breakdown (CAPEX):
| Component | Quantity | Unit cost (MXN) | Subtotal |
|---|---|---|---|
| 4K camera Hanwha XNV-9083R | 200 | $9,800 | $1,960,000 |
| 2MP camera Hanwha XNO-6010R | 30 | $4,200 | $126,000 |
| NVR Hanwha XRN-1620 | 14 | $32,000 | $448,000 |
| 24-port PoE+ switch | 14 | $8,500 | $119,000 |
| Cat6 cabling + installation | 14 stores | $90,000/store | $1,260,000 |
| Genetec Security Center + licenses | 230 licenses | $4,800/cam | $1,104,000 |
| Genetec central server | 1 | $185,000 | $185,000 |
| CAPEX subtotal | — | — | $5,202,000 MXN |
| 10% contingency | — | — | $520,000 |
| Total CAPEX | — | — | $5,722,000 MXN (~USD $286,000) |
Annual OPEX (year 1):
- Maintenance (Hanwha + Genetec support): ~$280,000 MXN/year
- Genetec licenses (includes support): $260,000 MXN/year
- Storage (disk replacement every 3 years, prorated): $80,000 MXN/year
- Year 1 OPEX: ~$620,000 MXN (~USD $31,000)
5-year TCO: CAPEX $5.7M + OPEX 5×$620k = $8.8M MXN (~USD $440,000).
Implementation (Q2-Q3 2025)
The project ran in 3 phases of 5 stores (not all at once, to keep each store operating during the upgrade):
Phase 1 — Pilot (5 stores): 6 weeks
- 1 week cabling in each store (overnight, without closing store)
- 2 days installation + commissioning per store
- 1 week parallel operation (new cameras active, legacy DVR as backup)
- 1 week legacy DVR decommissioning
Phase 2 — Rollout stores 6-10 (5 stores): 4 weeks
- Same pattern, trained team, faster
Phase 3 — Rollout stores 11-14 (4 stores): 3 weeks
- Maximum efficiency
Total downtime per store: 4 hours (during final commissioning, in 2-6 AM overnight window).
Results (12 months post-implementation, Q3 2026)
Shrink:
- Pre-upgrade (Q3 2024): 1.38% of sales
- Post-upgrade (Q3 2026): 1.12% of sales
- Reduction: 0.26 points = $4.7M MXN/year recovered
Incident investigation:
- Average time to find incident: 45 min → 8 min (82% reduction)
- Incidents resolved per month: 32 → 48 (those that were not resolved due to time cost now are)
- Corporate investigation person-hours: 120 → 35/month
Embedded analytics (Genetec KiwiVision + Hanwha AI Engine):
- Loitering detections in warehouse zone: 38/month (previously uncounted)
- Line crossing in restricted zone: 22/month
- Persons in liquor zone >30s: 14/month
- False alarms from motion: 8% (tuned threshold, was 35% in month 1)
Calculated ROI:
- Shrink savings: $4.7M MXN/year
- Investigation person-hour savings: 85 hours/month × $250 MXN/h = $255,000 MXN/year
- Better incident resolution savings: ~$600,000 MXN/year (estimate, based on reduced unclaimed losses)
- Estimated annual benefit: ~$5.5M MXN
Payback period: CAPEX $5.7M / annual benefit $5.5M = ~12.4 months.
The 3 things that went wrong
1. Analytics false alarms month 1 (35% of alerts were false) The Genetec KiwiVision system shipped from the box with too sensitive default thresholds. The first month we had 35% false alarms (outdoor vegetation movement, cloud shadows, animals). Solution: 2 weeks of zone-by-zone tuning, AI Engine detection calibration, schedule adjustment (do not alert between 2-5 AM in low-criticality zones). By month 2 it dropped to 12%, by month 3 to 8% sustained.
2. Hidden cabling cost in stores with already-built slabs 2 of the 14 stores had concrete slab, not drywall. Cabling had to go through exterior ducts (cable tray), which added $35,000 MXN per store not contemplated in the original CAPEX. Solution for future stores: architectural survey up front, cable tray planning from the design.
3. Store staff training We assumed the store manager would use Genetec Security Center without problem. Reality: 70% of managers had never used VMS, they relied on the night guard to “review cameras”. Solution: 4 hours of training per store, printed manual of “10 most common searches”, and a direct WhatsApp channel with corporate IT support. The training investment (~$45,000 MXN) was not in the original budget.
Lessons learned
- The ROI is not in the camera, it is in the search. The reduction from 45 min to 8 min in incident search is the most tangible operational benefit, not the image quality improvement.
- Centralized VMS > independent NVR. The ability to view any store from corporate, without asking the guard for footage, changed the operating model.
- Embedded analytics requires initial tuning. Do not install and let run. The first 2-4 weeks are calibration; budgeted as part of the project.
- Store staff training is not optional. Without it, the VMS becomes a DVR with better image.
Executive summary
For a Mexican retail chain of 10-20 stores, 800-1500 m² format, migration to 4K IP with centralized VMS pays back in 12-18 months via shrink reduction + operational efficiency. Typical CAPEX: $300,000-$400,000 USD for 200 cameras. Annual OPEX: ~10% of CAPEX. Payback: 12-18 months.
P.D. Does your chain have more than 10 stores with legacy DVRs? If yes, the migration ROI is in the 12-18 months range with real numbers — not in “3-5 years” as a vendor would tell you.
Frequently asked questions
How much does it cost to migrate 200 analog cameras to IP in a 14-store Mexican chain? Typical CAPEX: $300,000-$400,000 USD, including 4K IP cameras, NVRs, PoE+ switches, Cat6 cabling, centralized VMS, and labor. Component breakdown is usually 35% cameras, 25% NVR/storage, 15% networking, 15% installation, 10% VMS and configuration.
What’s the typical payback for a migration like this? 12-18 months in Mexican retail with 800-1500 m² format. Half the payback comes from shrink reduction (1.08% ANTAD → 0.65-0.75% post-VMS with analytics); the other half from operational efficiency (incident investigation 45 min → 8 min average).
Centralized VMS or independent NVR per store? Centralized VMS. The reason is not technical, it’s operational: the ability to view any store from corporate changed the management model. Corporate moved from “waiting for the store manager’s report” to “verifying in real time” — and that is only possible if all stores write to a VMS accessible via WAN.
How much calibration time do embedded analytics need after migration? 2-4 weeks per store. People counting, heatmaps and occupancy analytics need 1-2 weeks of “training” in each physical environment (light, angles, traffic patterns) to reduce false positives to <5%. Budget it as part of the project, not as an extra.
What if only some stores have good WAN — is centralized VMS still worth it? Yes, but with a hybrid model. Stores with good WAN write to the centralized VMS; those without write to local NVR with batch sync to corporate every 4-6 hours. 90% of the operational value (shrink reduction) stays in-store; the 10% (cross-store audit, real-time alerts) is what needs WAN.
Migrate all at once or in phases? In phases. 3-4 stores as pilot, 2 months of real operation, then rollout of 4-5 stores/month. Migrating all 14 at once multiplies operational risk: a centralized VMS outage on day 1 leaves you with no visibility of the entire chain.
Sources
[1] ANTAD — Merma retail México 2024: 1.08% sobre ventas — https://www.antad.net.mx/ [2] NRF — National Retail Security Survey 2024 (US benchmark) — https://nrf.com/research/national-retail-security-survey-2024 [3] Hanwha Vision — X-series 4K IP camera family datasheet — https://www.hanwhavision.com/en/products/network-cameras/x-series/ [4] Genetec — Security Center pricing and license model — https://www.genetec.com/products/security-center [5] Milestone Systems — XProtect VMS comparison — https://www.milestonesys.com/products/ [6] IFSEC Global — Analog to IP migration cost analysis — https://www.ifsecglobal.com/ [7] SANS Institute — Video surveillance network design best practices — https://www.sans.org/white-papers/