Too many people or too few: coffee shops and crowd density

Picture this: you're walking by a new coffee shop and pause to take a look. Hmm, how can you figure out whether they have good coffee - maybe the # of people inside? Too few customers and you might think "boy, this place probably sucks." Too many and you're no longer interested in fighting the crowds for just a cup of coffee. 

On the web we'd just multivariate test to find optimum page layout. But the darn physical world is so much more complicated. I threw together a short survey and farmed it out to 100 US consumers via mechanical turk (side note: using turk is addictive). All the usual disclaimers reply WRT survey bias and construction - i know my questions were a bit leading...

Imagine you were going to try a new coffee shop....


1. If the cafe was empty, would that make you more likely to enter, less likely or no change?

No Change: 52%

Less Likely: 26%

More Likely: 22%

2.  If this new cafe had good coffee, how many other customers would you expect to see?

Avg: 11.1 

> those who answered 'less likely' in Q1 avg'ed 14.5

> those who answered 'more likely' avg'ed 9.6

0 - 5: 30

6 - 10: 40

10 - 19: 12

20+: 18

3.  How many people would need to be in a cafe to make it "too crowded" for you? 

Avg 25.7

> those who answered 'less likely' in Q1 avg'ed 30.8

> those who answered 'more likely' avg'ed 22.9

0 - 10: 11

11 - 19: 23

20 - 29: 35

30+: 31

So it would suggest that the average cafe owner would want to have ~11 - 19 visible in their cafe to people on the street. Any more than that and you should conceal them in a back room or out of sight of the front windows and register :)