Case Study · Online Grocery · Dark Store Operations
Building the operating baseline that turned dark-store KPIs into a management system
A UAE online grocery operator with multiple dark stores had no unified model for labor productivity, capacity planning, or cross-role performance comparison. Vorctis built a dark-store operational baseline from the ground up — covering order volume, AOV, headcount, picker load, and driver load — then layered role-based KPI scorecards and governance rhythms on top. The result: measurably faster pick cycles, higher labor utilization, and a wrong-pick rate tracking toward operational target.
About the Customer
An online grocery operator running several dark stores across the UAE, managing high daily order volume, tight delivery windows, and multi-temperature picking, packing, and dispatch flows.
Problems
- Pick productivity varied significantly across stores, zones, and roles with no unified model to diagnose the gap.
- Labor planning and shift sizing lacked a baseline — no shared data model linked order volume to headcount requirements.
- Pickers, drivers, and store staff were measured by incompatible metrics, making cross-role performance comparison impossible.
- Order volume swings were not mapped to a headcount model, creating recurring capacity planning pressure.
- Wrong-pick rate sat above operational target with no systematic governance to drive it down.
Solutions
- Built a dark-store operational baseline model integrating daily order volume, average order value, headcount, picker load, and driver load per shift.
- Designed role-based KPI governance with scorecards for each operational function — picker, driver, and store staff — on a common measurement framework.
- Delivered operational dashboards surfacing daily throughput and labor-utilization signals for shift managers and leadership.
- Implemented performance-management governance logic linked to KPI scorecards, establishing clear review rhythms and accountability at each role level.
- Created a digital-twin style representation of the store operating model, enabling what-if capacity planning and baseline scenario comparison.
Impact
Impact
Several key metrics improved, with exact numbers used only when approved.
Order volume — baseline vs after governance
Weekly order volume held flat through the baseline period, then climbed steadily once the KPI operating model and review rhythm were in place.
All data is synthetic and directionally representative. No source figures, customer identifiers, or commercially sensitive values are embedded.
Our approach
How We Solved It
We measured the operation before we modeled it
Before any KPI or scorecard existed, we captured how the dark stores actually run — across stores, zones and shift patterns — using four complementary methods.
Process capture & mapping
We shadowed and mapped the real pick, pack, stage and dispatch flows end to end — documenting every hand-off, queue and exception, not the idealized process.
Operator & leadership interviews
Structured interviews with pickers, drivers and shift managers surfaced friction, workarounds and the tacit rules that never show up in the system logs.
Throughput measurement
Time-and-motion sampling combined with system-log analysis quantified items per labor hour, pick time per order, and wait/idle — the raw signals behind the KPIs.
Computer-vision capture
At control points — packing benches and staging / dispatch lanes — computer vision measured handling time, congestion and accuracy events continuously, without manual stopwatching.
One baseline operating model
We unified the captured signals into a single operating model that links order volume, AOV, headcount, picker load and driver load per shift — the shared baseline every role and decision is now measured against.
Inputs
Operating model
Decisions
The model became a management system
Role-based KPI scorecards, dashboards and a weekly review rhythm so the numbers drove coaching, staffing and dispatch — not just reporting. Labor utilization rose from 61% to 79% and pick productivity climbed past target trajectory.
Labor utilization — pickers & drivers
—
Pick productivity — items per labor hour (indexed)
Gains that compound
Wrong picks per 1000 SKUs
Ready to solve your operational problems?
Tell us about your operation and we will show what a similar improvement roadmap could look like.
Contact usContact
Contact Vorctis Digital
Share your operation context and we will come back with the right next step.
Request received
Your request was received.
Someone from Vorctis will be in touch shortly.