Engineer? Harder to say. Data scientist--that's easy. Stackbucks is very good at looking at demographic data and recognizing where future demand will be new new locations. They basically mark gentrification.
I used to work with a guy whose wife worked for Starbucks. He explained that they'll do seemingly crazy shit, like put stores across the street from each other, that makes sense when you look at the data. The store on one side of the street picks up traffic from point A to point B, while the other one gets point C to point D, optimizing for not making their customers walk across the street.
I think of it as similar to how UPS rigs their routes to avoid left turns.
Do you work in commercial real estate? My father does, and would pick them and Walgreens as the most effective scaled siting operations in the US.
Optimizing for absence of failed stores would be an interesting utility function for them to have, but since they own substantially all of the stores (and therefore don't have franchisee relationships to consider, where volatility for a partner might impact you disproportionately to economic loss), that would be a suboptimal utility function. They should calibrate their risk appetite to maximize profits; occasional failed stores are just a search cost.
I've seen a Walgreens that was closed in less than a year because of poor location despite being across the street from a Rite-Aid. They aren't perfect either.
Walgreens will happily close a location and move it across the street to take advantage of a more advantageously shaped curb cut if they can't get the Illinois Department of Transit planner to approve the optimal curb cut from their current location. This is because they have extreme confidence in their ability to assess the business impact of curb cuts.
A market commenter who believes that Google is not one of the most sophisticated engineering organizations in the world is wrong. If you believe Walgreens is not one of the most sophisticated commercial real estate operations that exist, I would encourage you to talk to people who understand real estate as well as you understand Google and recalibrate based on what they tell you.
I know of Walgreens locations that have been profitable for 50 years or more.
MacDonalds is similar. If they have a franchise that is doing badly, they will pay to move the franchise to a new location. A suburban McD closing is rare, usually only the oldest ones on lots that are too small for contemporary usage patterns.
Chipotle's former CEO (and I believe founder) often mentions that Chipotle's time under McD's ownership hammered home that the single most important element of the company's success was excellence in real estate. It's also why getting a McDonald's franchise is generally very expensive - they'll give you as de-risked a location as possible and a proven operational playbook. It's not easy to run a franchise at all (you have to be available on days your manager calls in sick, deal with vendors, inventory, etc), but you don't need to figure out the business model at all.
The point is that failed stores is not a good metric to optimize for.
You can have 0 failed stores if you open 0 stores. A successful operation will inevitably have failed stores. Having a long list of failed stores doesn't mean anything in of itself. You have to look at the rate of failed stores vs successful stores among other metrics.
It's not that useful to say they aren't perfect. The better measure would be how they compare to other stores, and/or how their own churn rate has changed over time.
This also misses the point. The point was that optimizing for number of stores closed is the wrong metric to begin with. Who cares how they rank on a metric that doesn't matter?
The original point wasn't that # of stores closed is an optimization problem, but that it was a valid way to judge the abilities of the people making the decisions.
And the counterpoint is the number of closed stores is not a good metric to judge the people making these decisions. In fact, judging the number of closed stores may well be the entirely wrong metric to judge them by.
Historically Starbucks has been used as a market indicator for other companies' expansion. If a starbucks opened in a location then it was assumed that there were growth factors at play in the market since that was the north star for Starbucks.
Starbucks in 2019 is the McDonald's of coffee shops, so it's not the same, but if you track their Starbucks Reserve locations you'll find the same predictions being used.
