One thing that Starbucks _has_ done is put a lot of thought into the flow of work behind the counter. When there are busy periods, staff go into more specialized functions; drinks cups aren’t walked from one location to another; the barista only does hot drinks; another does cold and brewed; etc., etc. They have already managed a lot of the variation induced by the variety by having the specific variety options (flavors; milks; sugars) in close proximity, so the combinatorial problem isn’t as bad: flavor / not flavor instead of which of 15 flavors. They have also managed a lot of variability in quality by installing more automated espresso machines that take out more of the “art” of making the espresso and steaming the milk.

Looking at other coffee shops with different perspectives on what they offer. High end coffee shops are SLOW because they put care and attention into each cup. They still have specialized positions, but they aren’t anywhere close to “production line” like Starbucks or Petes. Customers are willing to wait for what they believe is a better quality product.

Baristas also walk over empty cups to the drink maker. Seems like a walking waste. I wonder if remote printers could be used instead.

You were consistently clear on the assumptions in the model and warned early on that you were putting “brand identity issues and customer experience aside” but we must still revisit them when we get to the recommendations.

For Starbucks, following the first of the recommended steps needed to “establishing a consistent order fulfillment drumbeat” would be a mistake as it would seriously undermine its brand position. And I say this as a coffee drinking TOC guy who loves to see DBR at work.

Starbucks, like the local espresso stand, promises and is expected to deliver a high mix, user customized product. They made a choice to standardize not the product mix but their quality and service levels. Verifiable evidence of these choices include the high profile replacement of automatic espresso machines with older designs that allow more operator control as well as the elimination of the practice of staging shots for near future use.

Starbucks balanced these choices against optimizing takt, relying on their customer’s willingness to wait (both to order and to accept delivery) in return for receiving their exact drink of choice within a higher and reliable quality and service range. Starbucks knows they have an internal constraint, but usually only at peak demand, and it’s one they’re ok with as long as their customers are ok with it too.

Instead of Lean I think Starbucks is using a perturbation on an S-DBR approach.