Fast-movers make dark-store ops managers guess, every single day
Demand swings hard
Fast-moving items see sharp day-to-day swings, with occasional spike days assumption
Capacity gets caught out
In Oct 2025, Blinkit paused onboarding new products network-wide as festive demand outran warehouse capacity sourced
Fixed rules don't adapt
A static reorder rule doesn't move with demand variability, so the mismatch shows up store by store assumption
Problem statement: dark stores can't predict day-to-day demand swings precisely enough, so a fixed reorder rule either runs out of fast-moving stock or wastes shelf space holding too much of it.
More buffer stock cuts stockouts, but it isn't free
Holding more buffer stock of a fast-mover cuts stockout risk, but ties up shelf space that's already scarce at a dark store assumption and adds holding/spoilage cost sourced, general retail principle. This is the real lever the game below makes you feel.
Restock Run — set a reorder policy, watch 12 days play out
One SKU, one store, 12 days. Set how much safety buffer to carry and run it. The whole playthrough plays out on its own: no daily decisions, just watch the shelf, the deliveries, and the cost tally. At the end, your policy is compared against a benchmark reorder policy computed on the exact same demand.
Higher buffer → you reorder earlier and carry more spare stock: fewer stockouts, more holding cost. Lower buffer → cheaper to hold, riskier on spike days.
Holding cost: ₹2 per unit, per day left on the shelf.
Stockout cost: ₹8 per unit of demand you couldn't meet.
My recommendation: dark stores should size their reorder buffer per SKU using demand data i.e. how much it swings day to day, instead of one gut-feel or one-size-fits-all rule for everything. Play the simulation above to see exactly what that's worth in rupees.
Primary metric + guardrail, from your run
The primary metric is total cost per SKU over the run i.e. holding cost and stockout cost combined, lower is better. The guardrail is stockout rate: a policy that hits a low ₹ number by chronically underordering should get flagged here, not rewarded.
Why this matters for Store Ops
A few rupees of holding cost per unit sound trivial next to a stockout on a top-selling SKU — until you multiply by thousands of dark stores and hundreds of fast-moving SKUs each. The reorder policy a store runs, not just the demand it sees, is what decides which side of that trade-off it lands on. That's a Store Ops / Supply Chain PM problem, and it's the kind of trade-off I like digging into.