Fresh meat is the cruellest category in e-commerce. Customers expect store-bought-quality cuts at their door inside a two-hour slot, in 42°C summers, with batch labels they can read. Inventory has a shelf life measured in days. A missed expiry isn't just lost margin — it's an unhappy customer and a regulatory paper trail. Licious has run this gauntlet for ten years across twelve Indian cities, and last year they rebuilt the warehouse and logistics stack on Polluxa.
We sat down with Madhura D., their head of supply chain, to ask what actually changed.
The starting state
Polluxa: Take us back. What did the old setup look like?
Madhura: We were running three different systems — one for WMS, one for cold-chain routing, one custom-built for batch and expiry. The three did not talk in real-time. So at the end of every day, someone in finance ran a reconciliation script. About 4% of our daily SKU-units ended up in the "where did this go" column. That's not great when you're selling fresh chicken.
The bigger issue was FEFO. First-expiry-first-out is the only sane picking strategy in fresh, and we were running it manually. The picker would look at the bin, look at the dates, and pick the shortest one. Mostly. Some days, when the order was big and the floor was busy, they wouldn't. So we'd ship a customer a cut with three days left when a one-day cut was available. The one-day cut would spoil. The customer who got the three-day cut would be fine. We were losing the math both ways.
What changed
Polluxa: What's different now?
Madhura: Three things, mostly. First, FEFO is baked into the picking app. The picker scans the bin and the scanner tells them which specific unit to pull. They don't have to look at dates. They don't have to think. Spoilage from picking errors went to zero in eight weeks.
Second, we have one inventory number across our app, our quick-commerce partners, our gifting flow and our enterprise B2B. Before, our app would say "in stock" while Zepto would say "sold out." Now it's one truth. No oversells, no underselling.
Third — and this is the one I tell other operators about — the cold-chain agent. It watches every parcel from pickup to delivery. If a rider's bag temperature drifts above the spec, even briefly, the order is flagged before the customer opens the package. We've caught equipment failures three days before our normal monthly maintenance would have found them. It's quietly the most useful agent we have.
FEFO is not a feature you can ship without changing how the picker behaves. You have to remove the choice. The picker should not be deciding which unit to pull. The scanner should.
Madhura D., Head of Supply ChainThe numbers
Polluxa: Numbers tell the rest of the story.
Madhura: Spoilage dropped 38%, year-over-year, comparable category mix. We had assumed the floor was 5% — that's what every fresh operator I'd talked to ran at. We're at 2.8% now, and our analytics team thinks 2% is achievable. FEFO accuracy is 99.4% (up from a generous 84%). And we've cut our daily reconciliation time from 90 minutes to about 8.
The thing I didn't expect: complaints fell. Not because spoilage drove most of them — it didn't — but because the cold-chain agent catches the edge cases that used to land in customer support inboxes after the customer had already had a bad day. We are reaching out to customers about temperature events before they reach out to us. NPS is up 11 points.
What you'd do differently
Polluxa: What would you tell your previous self?
Madhura: Move on the agent layer first, the WMS migration second. We did it the other way around. The WMS go-live was a six-week project. The cold-chain agent we turned on in a day, and within that day it told us things our humans had been missing for years. If I were doing it again I'd start with the agent, get the visibility into our own ops, then migrate the system of record afterwards. We had assumed we needed to migrate first to "give the agent good data." Wrong. The agent reads real signals — sensor temperatures, GPS — that didn't need any migration.
One more thing
Polluxa: Anything you'd want other F&B operators to take from this?
Madhura: Two things. One: if your floor is debating which unit to pick, you don't have FEFO. You have FEFO-shaped good intentions. Two: temperature events are not rare. They're constant. You just can't see them. Once you can, you can fix them. Once you fix them, your customers stay.
Read the full Licious case study for the metrics breakdown, or explore Polluxa WMS to see the FEFO engine in action.