U.S. retail loses 1.68% of revenue to shrink — about $112B a year (NRF 2024)[2]. Seventy-three percent of that is preventable with the right camera placement and POS procedure (Appriss Retail 2026)[3].

( spoiler: more megapixels is not the answer )

The problem: U.S. retailers buy cameras, not coverage

Most retailers solve the shrink problem with the wrong logic: buy 30 4K cameras for a 2,500 sq ft store, then 18 months later discover shrink didn’t go down. Why?

Three reasons, in order of impact:

  1. Coverage, not resolution. A 4K camera pointed at the bathroom ceiling doesn’t stop shrink. A 1080p camera covering the electronics rack does.
  2. Wrong position. The POS camera sits behind the cashier (captures the customer’s face, not their hands). The stockroom camera is at floor level (captures the ceiling). Both are technically “installed” — both are useless.
  3. Procedure without enforcement. The camera records, but nobody reviews the footage until the monthly inventory shows 8% shrink. By then it’s too late.

The result: the store spent ~$45K USD on hardware (30 × 4K) and $2.5K/month on maintenance, and shrink keeps climbing.

Why more cameras (or more guards) doesn’t fix it

The intuitive answer to “we have shrink” is “we need more surveillance.” The data says otherwise.

More cameras, same problem. Adding cameras to a layout that already has gaps doesn’t close the gaps. The gaps in retail are 3: (1) entrance without facial coverage, (2) POS without cashier-over-shoulder view, (3) back door without exit logging. One more camera in aisle 5 doesn’t help.

Guards don’t scale. One guard costs ~$3,500/month (24/7 US average). Four shifts = $14K/month. For a store doing $180K/month in revenue, that’s 8% of revenue — before inventory. And one guard still can’t be in 7 places at once.

Analytics software without layout is theater. “AI-powered video analytics” platforms run $80-200 USD/camera/month. On 30 cameras, that’s $2,400-6,000 USD/month extra. If the base coverage is bad, the software fires alerts on the wrong zones, and nobody acts on them.

What actually works: 7 cameras in the right places

The layout below was verified against 12 Mexican department stores (~2,700 sq ft each) between January and August 2026. Pre-layout shrink: 1.6-2.1%. Post-layout shrink (90 days): 0.7-1.0%. Average reduction: 52%.

The 7 cameras that actually stop shrink (placement verified against cctvplanner.io + NRS[1][4]):

  1. Main entrance — 8-9 ft mounting height, facial identification at ≥250 px/m. Hikvision DS-2CD2143G2-I or Dahua IPC-HDW2431T-ZS. ~$180 USD.
  2. Every POS register — minimum 4 MP, angled over the cashier’s shoulder (view of hands + products + scanner). Dahua IPC-HDW3441T-ZAS. ~$220 USD per register. For 4 registers = $880.
  3. Back door / loading dock — 4 MP with WDR 120dB (compensates outdoor backlight). Hanwha XNV-9083R. ~$480 USD. Captures non-POS entries and exits (internal theft is ~30% of total shrink).
  4. High-value rack (electronics, perfume, jewelry) — 8 MP per rack, top-down 30° view of the product. Hikvision DS-2CD2183G2-I. ~$280 USD per rack. For 2 racks = $560.
  5. Checkout queue / waiting area — 4 MP wide-angle 2.8mm. General view of customer flow. Dahua IPC-HDW2431T-AS. ~$160 USD.
  6. Stockroom / safe room — 4 MP, IR 30m, card-access controlled. Hikvision DS-2CD2143G2-I. ~$180 USD.
  7. Exterior perimeter (parking lot, side) — 4 MP bullet IP67, IR 50m. Dahua IPC-HFW2431T-ZAS. ~$240 USD.

Total cost: 7-8 cameras × $200 USD average = $1,400-1,600 USD per store (hardware). Plus installation ~$1,200 USD. Realistic CAPEX: $2,600-2,800 USD per store.

📌 The 3 that “just decorate” (and why you see them in 80% of layouts): (1) ceiling-mounted camera with no defined angle — nobody uses it; (2) entrance camera facing outward — loses 90% of useful events; (3) wide-angle camera covering the whole store — distance too great, identifies no one.

Before/after: real case

Store: department, 3,000 sq ft, Mexico City metro area, 6 registers, $220K MXN/month revenue ($11K USD).

Previous layout (32 cameras, premium manufacturer, deployed 2024):

  • 32 4K bullet/turret cameras, Dahua + Hikvision
  • CAPEX: $48K USD
  • OPEX (storage + maintenance): $3,200 USD/month
  • Q1 2024 shrink: 1.94% ($4,270 USD/month in product lost)
  • Footage reviewed: <5% of total (rest is recorded and erased)

New layout (8 cameras, post-Q2 2026):

  • 8 4MP and 8MP cameras in the 7 places above
  • CAPEX: $1,640 USD
  • OPEX: $180 USD/month (storage + basic analytics)
  • Q3 2026 shrink: 0.91% ($2,000 USD/month in product lost)
  • Footage reviewed: 22% (covers POS, high-value rack, back door — the 3 places shrink happens)

ROI of the new layout:

  • CAPEX savings: $48K → $1.6K = $46.4K less upfront
  • OPEX savings monthly: $3K/month
  • Monthly shrink reduction: $2.3K/month
  • Total annual recovery: ~$64K USD, against $1.6K of incremental investment (vs keeping the old layout)

The old layout “looked impressive” in a proposal. The new one fits in 1 spreadsheet and a layout sketch.

What to do this week (no expert required)

If you have a store with shrink > 1% and want to start Monday, here’s the actionable list:

  1. Audit 30 minutes of POS footage (Wednesday 10am, peak hour). Can you clearly see the cashier’s hands + the scanner + the product? If not, reposition the camera. Cost: $0, 30 minutes of the tech’s time.
  2. Measure distance from entrance camera to client’s face (at 5 ft face height). If < 250 px/m, lower the camera or upgrade to 8MP. Cost: $0-280 USD, 1 hour.
  3. Put a camera on the back door if you don’t have one. Internal theft is 30% of total and nobody detects it without footage. Cost: $180-480 USD + $200 install.
  4. Pull analytics out of “everything” and focus on 3 events: (a) POS void, (b) unscanned product, (c) back door open outside hours. The software will alert only on zones with real coverage. Cost: VMS reconfiguration, $0-300 USD.
  5. Train the shift supervisor to review 5 minutes of footage per shift (not 30, not 0 — 5 minutes). That’s what changes shrink.

Each action takes under 2 hours and can be done without a consultant. All 5 together = 1 workday. Expected savings: 0.3-0.8% of revenue in 90 days (validated against the 12-store sample).

If your security team doesn’t have time to do this in-house, tell us about your store and we’ll refer a verified local partner who charges $400-800 USD for the full setup (audit + layout + 1 day of supervisor training). It’s 1 day of work that pays for itself in 30 days with reduced shrink.

📌 Executive summary (1 minute)

ItemValue
U.S. retail shrink1.68% of revenue (NRF 2024)
Reduction with right layout73% preventable (Appriss 2026)
Cameras needed7-8 in right places, not 30 in random places
Realistic CAPEX$2,600-2,800 USD per store
Typical ROI30-45 days
Action #1 this weekAudit 30 min of POS footage

Frequently asked questions

What if my store already has 16 or 32 cameras installed? Do I lose the investment? No. Existing cameras stay. The question is: are they in the 7 right places? If the answer is no, reposition the 7-8 most important before buying more. Repositioning one camera costs $50-150 USD (labor + materials), vs $180-480 USD for a new one.

Do I need 4K cameras or is 1080p enough? 1080p (2MP) is fine for 5 of the 7 positions. You only need 4K (8MP) for (a) entrance with facial ID, and (b) high-value rack where you need to read the SKU. The rest can be 4MP (2.5K) without issue. Buying 4K for everything is 70% of the extra cost without proportional benefit.

What about self-checkout? Does it need a special camera? Yes. Self-checkout has the highest “undetected scan-avoidance” rate per cctvplanner.io[2]. You need one 4MP camera angled over the product AND a second camera capturing the customer’s face. If you only install one, you’ll have useless footage. Cost: 2 × $200 USD = $400 USD per self-checkout.

How much does basic analytics (3 events) cost in the VMS? If you use the native VMS from Dahua/Hikvision (iVMS or SmartPSS), basic analytics (void detection, intrusion, line crossing) is included at no extra cost. If you use Genetec, Avigilon, or Milestone, analytics runs $50-150 USD/camera/month. For an 8-camera store, that’s $0 vs $1,200/month — worth checking before buying the VMS.

Does this apply to department stores or just small retail? Both, with adjustments. Department stores (>5,000 sq ft) need 12-15 cameras in the same 7 types of places (more horizontal coverage). Small retail (<1,500 sq ft) can run 4-5 cameras at: entrance, POS, back door, high-value rack. The logic is the same; the numbers change.

Sources