While watching an episode of Top Gear the other day (one of the best shows on the air these days, IMHO), I was greatly entertained by their trip to India. One segment really jumped out. There is a lunch delivery service that operates in Mumbai, operated by workers called dabbawala (literally meaning "one who carries a box").
A fleet of workers pick up hot lunches (all in paint can-style metal cans) from households throughout the metro area. They transport the lunches to the city, and then deliver them to the worksites at lunchtime. After lunch, the cans are delivered back home again.
Volume, Volume, Volume
What sort of volume do they handle? It's not a small operation. Approximately 5,000 people deliver over 200,000 lunches every day. That's a lot of cans. How do you organize and run such an operation? Rigorous closed-loop processes with auditability and no room for error? Do they use sophisticated systems to track each can and route it to the appropriate location?
Actually, no. A destination is marked on the lid of the can with color and some short codes. All the cans are thrown in carts and trains and carried around on shoulders and backs for delivery. There are lots of hand-offs from the cooking to the pickup to the transport into the city to the delivery. Yet they just all get there. :-)
There must be some mistakes and lost lunches, with such a loose operation. And yes, there are mistakes on occasion. But estimates are that they achieve Six Sigma delivery accuracy. (According to many sources; see notes below.) Which is a good thing, because how would you like to be one guy in the office who all of a sudden has to go buy a lunch? Embarrassing.
Lessons for BPM
What can we take from this? Automation isn't necessary for high-quality operations. Some things can be low-tech and old-fashioned but still cost-effective and high-quality. Also, don't underestimate the dedication and "corporate culture," if you will. Congrats to the dabbawalas!
The dabbawalas have been reported on many times over the years, I found. Dozens of articles reference a Six Sigma attribution from Forbes Global magazine in 1998 (also reported as being from 2002). Also, there are many references to Six Sigma meaning 99.999999% accuracy (ugh! did somebody think six sigmas means 6-nines?). There are references to the dabbawalas having one error in 6 million lunches, while other articles say of one error in 16 million lunches. My "favorite" is this quote from the Guardian:
"Forbes awarded the humble dabba-wallahs a 6 Sigma performance rating, a term used in quality assurance if the percentage of correctness is 99.9999999 or more. In other words, for every six million tiffins delivered, only one fails to arrive."
None of the three "facts" agree with each other!
Altogether, I could find no proof or original reporting of the dabbawala's statistics. Old Forbes Global articles aren't available anywhere I could find. (I did find an ISO 9000:2000 certificate from 2006.) It appears that one error in six million *may* be from an original estimate, at some time or another, but it's uncertain.