The buns were getting cold and I was on basement level three of an apartment complex in Yelahanka, looking for a service elevator that the security guard had described using landmarks that no longer existed. “Past the old generator room, sir. Left at the water tank.” There was no water tank. There may never have been a water tank.
This is Friday afternoon at FIKA B’LORE. Not the Instagram version, not the Shopify storefront with the warm photography and the clean copy about Swedish fika rituals. The real version. The version where two founders, a delivery partner, and a box of cardamom buns navigate the physical infrastructure of Indian residential life.
I had not anticipated that the hardest part of selling premium pastries would be the last 50 meters.
The supply chain for a ₹195 cinnamon bun across a Bangalore apartment complex is, per unit of distance, significantly more complex than anything I managed at Tetra Pak.
The Geography of Delivery
In my supply chain career, I spent years thinking about distribution networks. How to move toothbrushes from a factory in Malmö to a warehouse in Melbourne. How to optimize truckload fill rates across European logistics corridors. How to structure a 3PL relationship in Australia. Big distances, big volumes, big systems.
None of that prepared me for Prestige Shantiniketan.
A typical delivery run looks like this: 10 communities, 15-25 stops, spanning a 30-kilometer arc across North Bangalore. The route itself is manageable. Google Maps handles it, more or less, with the usual Bangalore caveat that a “15-minute drive” means “15 minutes plus whatever the traffic gods decree.” The problem is not between the stops. The problem is at each stop.
Every gated community in Bangalore is a small fortress with its own entry protocols, its own parking logic, its own relationship between the security team and the concept of “delivery.” Some complexes have a designated delivery area near the gate. Some require you to park in the visitor section on B3 and walk to a service elevator that may or may not be working. Some have a system where you call the resident, the resident calls the security desk, the security desk calls the gate, and the gate lets you through, a process that takes somewhere between three and twelve minutes depending on how many people are on lunch break.
Multiply that by 20 stops and you understand why a delivery run that covers 30 kilometers takes five hours.
What Supply Chain Theory Misses
In supply chain management, we have elegant models for the “last mile.” Plenty of companies have built billion-dollar businesses on solving last-mile delivery. But the last mile assumes you can get to the door. In Bangalore’s residential landscape, the problem is the last meter: the distance between the apartment complex gate and the customer’s kitchen counter.
This is not a problem that technology solves. It is not a problem that a better route optimization algorithm addresses. It is a problem that lives in the physical world, in the relationship between a building’s security protocols and a small business’s need to deliver perishable goods while they are still warm. The solution, insofar as there is one, is human: learning each complex’s quirks, building relationships with security teams, knowing which elevator to use and which entrance to avoid at 3pm when the school buses are coming in.
I spent seven years at Tetra Pak thinking about how to move packaging materials across continents. The supply chain for a ₹195 cinnamon bun across a Bangalore apartment complex is, per unit of distance, significantly more complex.
The Scaling Question
Right now, Niklas and I do most deliveries ourselves, sometimes with help from Porter or Borzo for the outer routes. This is deliberate. Not because we enjoy parking garages, but because doing deliveries yourself teaches you things that no dashboard ever will. You learn which communities have the highest reorder rates (the ones where the security guard knows you by name). You learn which time slots work (before 4pm, after school pickup, never during the lunch hour changeover). You learn what customers actually say when they open the box, which is information worth more than any NPS survey.
But it does not scale. Two founders cannot deliver 500 orders. At some point, probably soon, we need a delivery system that preserves the quality of the experience without requiring the founders to navigate every basement parking structure in North Bangalore. That is the problem I am thinking about most right now, and I do not have the answer yet.
The temptation is to outsource it entirely. Hand it to a logistics partner, give them the addresses, let them figure out the service elevators. But the handoff between “founder delivery” and “outsourced delivery” is exactly the kind of gap where quality falls through. The customer who orders because she knows Dennis will show up with a warm box and a smile is not the same customer who orders from an anonymous delivery service. The relationship is part of the product.
So the question becomes: how do you encode the things you learn from doing deliveries yourself into a system that someone else can execute? How do you transfer the knowledge that “at Embassy Springs, use Gate 2 and tell the guard you are going to Tower D” without losing the warmth that comes from the founder showing up? How do you scale a handoff?
I spent fifteen years in supply chain management, and this might be the most interesting handoff problem I have encountered. It is certainly the one where the stakes feel most personal. Nobody at Tetra Pak ever looked disappointed when a pallet of packaging arrived 15 minutes late. But a customer who expected warm cardamom buns and got lukewarm ones? That is a different kind of failure.
I am still working on it. I will let you know what I figure out.