Dark store video analytics for pick-pack SLA
· Updated · Slinai
Key facts
| Primary SLA | Order-to-dispatch: pick → pack → bay handoff, timed on camera + OMS |
|---|---|
| Productivity | Picker pathing, zone dwell, idle loops, bench congestion |
| First cameras | Worst pick cluster, pack table, rider bay |
| Hub | Dark stores video analytics industry page |
| Sibling industry | Cloud kitchens if the site is cook-and-bag rather than grocery pick |
Quick commerce SLAs die in physical minutes the OMS never photographed: a picker looping an aisle, a pack bench with three orders open, a rider waiting while someone searches for a bag.
Dark store video analytics is AI video analytics for that chain. Slinai runs on existing CCTV. The full capability list is on dark stores video analytics. This spoke is pick-pack productivity and order-to-dispatch SLA — not a second homepage.
Why OMS timestamps are not enough
OMS knows state changes. It does not know:
- Two pickers in the same narrow aisle.
- A packer waiting on a missing cold SKU.
- A rider bay that became a lounge.
Operations & productivity is the use case. The method is the same as retail labour: see the work, then change the floor — not a weekly lecture.
Join OMS order IDs when you can (data & integrations) so a clip is “that 10-minute order,” not “someone in a yellow vest.”
The three-camera starter kit
| Camera | Job | Alert worth sending |
|---|---|---|
| Pick cluster | Pathing, dwell, congestion | Aisle dwell over X minutes on live orders |
| Pack bench | Open orders, idle hands | Bench occupied, no bag-out |
| Rider bay | Handoff dwell | Rider wait over SLA |
Do not wire 40 rack cameras on week one. Restaurant kitchen SOP failed the same way: too many polygons, nobody to walk.
Productivity without turning people into scores
Publish team and zone views before individual heatmaps. Individual scores on bad camera coverage become HR incidents.
Use video for:
- Coaching a looping path.
- Rebalancing a hot aisle.
- Proving a layout change (same idea as retail conversion A/B).
Use WMS for units per labour hour. Argue less.
Shrink and hygiene sit next door
This spoke is SLA. Cluster siblings: order accuracy, rider-bay handoff, shrink / after hours, cold chain and FEFO. Do not mix “picker too slow” alerts with “theft” alerts in the same WhatsApp group.
How this ties to Slinai
- Hub: dark stores
- Throughput: operations & productivity
- Questions: agentic AI (“which dark store missed dispatch SLA at dinner?”)
- Deploy: existing CCTV
Book a demo with one late-SLA store’s pack and bay cameras. Bring yesterday’s OMS late-order list so the clips have names.
Common questions
Is this the same as warehouse WMS labour standards?
complementary. WMS knows order state. Video knows physical bottlenecks and handoff dwell the WMS never sees. Use both; do not replace WMS timestamps with pixels alone.
Do I need a camera in every aisle?
Start with pick face clusters that already miss SLA, the pack bench, and the rider bay. Add aisles when a supervisor will act on that polygon.
Can it measure order accuracy?
It can flag rushed pack behaviour and bay congestion; item-level accuracy still needs scan/WMS. Video is the exception and coaching layer.
What about shrink in a dark store?
Same platform, different playbook — high-value cages, exit, after hours. See the dark stores hub and the retail loss prevention guide for patterns.
Cloud or edge?
Many dark stores have usable WAN; some do not. Use the existing-CCTV guide: hybrid is common — local inference, NVR keep, events to HQ.
Related pages
Guides in this cluster point to live product, industry and use-case pages.
Related guides
Dark store order accuracy and defect prevention
Dark store order-accuracy analytics flags rushed pack, skipped QC and process drift on existing CCTV — a complement to WMS scans, not a replacement.
Dark store rider-bay handoff: the last 90 seconds of SLA
Dark store rider-bay analytics times queue depth, unattended bags and delayed handoffs on dispatch CCTV — the last 90 seconds after pick is already on time.
Cold chain and FEFO rotation on dark store cameras
Dark store cold-chain and FEFO analytics flags door-open dwell and shelf-rotation drift on existing CCTV — process evidence, not a logger replacement.