The Mumbai Dabbawalas: Precision Without Software Precision Without Software

A lunchbox delivery network in Mumbai runs on marks and handoffs. Here is how it works, and why I am careful with its famous error rate.

Every working day in Mumbai, a steel lunchbox leaves a home kitchen in the morning and lands on an office desk a few hours later. The person eating it never scanned anything. No driver was tracking a screen. So how does a hot lunch find the right desk in a city of millions?

The answer is a group of people called dabbawalas, and I have been reading about how they work. I think their system has a lot to teach anyone who runs a team, a shop, or a side project.

Who the dabbawalas are

The word "dabbawala" roughly means "box person", someone who carries a box. They collect home-cooked lunch in tiffin boxes (stacked steel containers) from homes across Mumbai, carry them to offices, and then bring the empty boxes back in the afternoon.

According to the Wikipedia article on the service, it began around 1890. A man named Mahadeo Havaji Bachche started a lunch delivery service in what was then called Bombay, with about a hundred men. That makes it more than a century old.

The same article says that roughly 4,500 to 5,000 dabbawalas move somewhere between 175,000 and 200,000 lunchboxes a day. Those are estimates from different reports, so treat them as a rough size, not an exact count. The article says the workers are self-employed. It also describes a charitable trust and an association, the second of which helps with legal issues and disputes. In other words, it is a network of independent people who agreed to follow one

The code on the lid

Here is the part I find most interesting. The article says each box carries several marks. There are abbreviations for the collection point, a colour for the starting station and a number for the destination station. There are also markings for the dabbawala who handles it at the other end, and for the building and floor. (The article is a little unclear on exactly what each colour means, so I will not try to explain more than that.)

Why marks and not names or addresses? The article says the average schooling of the workers is about the eighth grade, so the system uses colours and symbols that anyone can read quickly. That is the clever bit. The designers did not ask people to become something they were not. They fitted the system to the people who would use it.

The journey, as the article describes it, goes like this. A dabbawala, usually on a bicycle, collects boxes from homes. The boxes are taken to a sorting place and grouped, then loaded onto train cars marked for the unloading station. At the other end, a local dabbawala takes each box on to the right building. The empty boxes then travel back, after lunch or the next day.

Trust is part of the machinery

A system like this cannot run on marks alone. It needs people who trust each other with someone else's lunch, every day. The Wikipedia article gives a few clues about where that trust comes from.

It says dabbawalas typically start young, between about 15 and 20. Many come from the same small village near Pune, a city in the same state. A charitable trust was registered in 1956 and a commercial association in 1968. The article also says conflicts are settled in front of 20 foremen (called mukadams), who are selected every six years.

I would be careful about reading too much into this. I am working from one summary, and I cannot say how every detail works day to day. But the general shape is familiar. Shared roots, clear rules and a body that settles disputes make it easier for strangers to hand work to each other.

The article also records how the world noticed them. They were invited to the 2005 wedding of Prince Charles and Camilla Parker Bowles. In 2011 they went on strike for the first time in 120 years, to support a rally against corruption. The article also cites the New York Times saying the service was growing five to ten percent a year. A system that old, and reportedly still growing, is worth studying for that alone.

The famous number, and why I am careful with it

You may have seen a claim that dabbawalas make less than one mistake in six million deliveries. It is quoted in business talks everywhere. I wanted to know where it came from.

The same Wikipedia article gives the story. It says the number traces back to a 1998 Forbes article. The president of the local tiffinmen's association told the writer that mistakes happen "almost never, maybe once every two months". The writer then worked that out into a rate of one mistake in several million deliveries, and the article calls the result an estimate, not a careful study. It also says Forbes never certified the dabbawalas under "Six

So the safe version is this: the dabbawalas are widely admired for being very reliable, and Harvard Business School wrote a case study about them in 2010. But the neat statistic is an estimate that grew in the retelling. I would rather say that plainly than repeat a number I cannot back up.

I find that useful on its own. Good stories get shinier each time they are told. Before I repeat a number to a client or an audience, I want to know who measured it.

Why a simple system lasts

Reading about this, I notice a few things that hold up even if you ignore the exact numbers.

The system fits its people. Marks that need no reading skill work for everyone, so no one is left out when a new person joins.

Each person does one small job. Nobody carries a box from start to finish in one go. Each handoff is short and clear, so a mistake in one place is easy to spot.

The rules belong to the group. The article describes foremen who settle disputes, and says the association was reportedly started after a customer treated a dabbawala badly. People tend to follow rules they helped build.

It keeps going through disruption. The article mentions that the service pauses for five days each March for a village festival. A system that plans its rest days is a system that expects to last.

None of this needs software. That does not mean software is bad. It means that if a process only works when the tool works, it is fragile. If it works on paper first, a tool can make it faster later.

What I take from this

When I look at a process, these are the three plain questions I try to ask. Can a new person understand the steps in a few minutes? Is each handoff small and clear? And if the phone and laptop died today, would the process still run? A process that fails the third question is worth knowing about.

One thing to try this week: pick one repeating job in your business, such as sending an order out or onboarding a client. Write it as five or six steps, each one clear enough to hand to a stranger. Then ask who would get stuck, and fix that step first.

Sources

• Wikipedia, "Dabbawala": https://en.wikipedia.org/wiki/Dabbawala