Image

For my lunch @ work, I get it from VITAL foods, an agency which uses the famous dabbawalla network to deliver. Most of us are aware of the accurate service of dabbawalla’s in Mumbai. I think it is also a case study at HBS. Also, the publicity it gets in the local newspapers makes sure that most of the non-residents of mumbai also know about it.

However there was a letter from VITAL foods stating that they would discontinue the services of Dabbawalla as there have been too many complaints in the recent past. Being a customer of this system, I did find that to be true. Even though the system was accurate in the delivery , it was erratic in terms of the time of delivery.

For any process , time and accuracy both play a crucial role. It looks like dabbawalla score 6 sigma on accuracy of delivery, but score very poorly on the delivery itself, meaning the time to delivery.

I remember , well, it was ages ago , when I was trying to improve a process in GE where there were about 20 people doing some basic outsourcing transaction driven process . The team was great on accuracy but sucked on turn around time. There is actually a subtle trade off, or, it might seem that way , between accuracy and turn around time(TAT) for a process. Spent about 2 months trying to improve TAT for the process. Did some jazzy stats, build a small tool , to improve the TAT for the process and realized that TAT sigma went up but accuracy sigma went down. But over all client was satisfied and didn’t care about the sigma the team was hitting :) .Well, that my experience from one project.May be there are ways to take care of both the metrics.

May be Dabbawallas will figure out a way to take care of both the metrics, some day !.

In this context I think, 5 to 6 sigma on the relevant metrics(TAT and accuracy) is more important than maintaining 6 sigma on a subset of those metrics(Accuracy alone).