The Starbucks empire

This weekend, after encountering a triumvirate of Starbucks stores in a visit to Los Angeles, I wanted to analyze and visualize the company's global distribution. This topic was already explored in this (excellent) 2014 QZ article, but the Starbucks empire has now had another decade to grow, and hence deserves a new analysis.

I used the Starbucks store locator to acquire a list of 40606 stores.¹ The resulting data includes each store's latitude/longitude, address, and other fun information.

Loading...

By country

Here are 85 countries ordered by total number of Starbucks:

And here are the top countries by number of Starbucks per capita:

Well-endowed cities

Shanghai has by far the most Starbucks, at 1134, followed by Beijing (590), Hangzhou (466), Shenzhen (364), Guangzhou (332), and Suzhou (297). The first non-Chinese city is Istanbul (277), and the first US city is Chicago (194), followed closely by NYC (191).

Well-endowed US states

And per capita:

(Thank God for Mississippi....)

Isolated points

In the aforementioned 2014 article, the point on Earth furthest from a Starbucks was "just off the coast of South Africa", but Starbucks has now expanded there. Thus, the new "Point Nemo" of Starbucks is at 83° 36′ S, 19° 45′ E, which is in Antarctica. This point is 3395 mi (5464 km) from the nearest Starbucks. (I would not want to live there!)

Owing to the construction of a Starbucks in Glasgow, Montana, the most isolated point in the contiguous United States is no longer in Montana (as it was in 2014), but at 38° 33′ 36″ N, 115° 56′ 06″ W, in the Nevada desert just off Highway 6—157 mi (253 km) from the nearest Starbucks. (I would not want to live there, either!)

Visiting every Starbucks

I threw the LKH traveling salesman problem (TSP) solver at the list of 40606 Starbucks and found the following path to visit all Starbucks in a loop, with total length about 411000 km:

Loading...

(The lines sometimes look curved because they are great circles, while the map is a Mercator projection. Furthermore, lines are clipped when they cross the antimeridian.)

The solver did a great job at a large scale, but individual cities are usually poorly optimized. I suspect I need to break up the problem into subproblems to get a better solution.

Extremal Starbucks

The northernmost Starbucks is located in a shopping mall in Bodø, Norway, with coordinates 67° 16′ 39″ N, 14° 25′ 23″ E. The southernmost Starbucks is located in Invercargill, New Zealand, at 46° 24′ 43″ S, 168° 20′ 54″ E.

Most isolated Starbucks

The most isolated Starbucks is located in a Safeway in Thompson, Manitoba:

The two closest Starbucks

I've always found it amusing how close two Starbucks locations can be. At 32° 51′ 58″ N, 117° 13′ 56″ W—just south of UCSD—there's one at an entrance to Ralphs Fresh Fare, literally one door over from a standalone Starbucks. The most extreme cases are of this type: an independent store next to one ensconced in a grocery store.

Unfortunately, the coordinates are not terribly accurate, especially for non-independent stores, so finding a definitive answer here will require manual analysis of several hundred pairs of stores within 100 meters of each other. We can, however, find the two closest independent stores!

After filtering out weird bits of data like Starbucks located on airport runways (??), the closest two independent Starbucks appear to be in Paris (Google Maps image below), 82 meters apart:

I'm nowhere near certain of this answer, though, because a lot of the coordinate data is low quality.

Further work

• The data need to be cleaned up—I'd estimate there are about 250 duplicates.
• I'd like to find minimal circles that enclose n Starbucks.
• I want to make a better interactive map.

I won't be making the data and code public because it's hella messy, but I will share it on request; just shoot me an email.