The statement that “Nobody goes there anymore. It’s too crowded.” is usually attributed to Yogi Berra. But it raises an interesting question: How do you know if it is crowded if you don’t go?
As the Globe and Mail reports, a British software developer is taking on this question by mining Foursquare data (FourSquare, the anti-social app?, Feb 9).
As a growing number of people flock to Foursquare destinations, When Should I Visit shares the best time to visit various attractions. The definition of “best time,” in this case, is the least busy time. For example, if Tuesday is the most popular check-in day at The British Library, you might want to wait until the slowest check-in day to drop by (which, according to Foursquare data, is Sunday).
Here are some examples of what When I Should I Visit tells you:
I think it hardly surprising that the British Museum is busier on the weekend than during the week. Indeed, you might guess that would be true for most cultural attractions but as the Institute of Contemporary Arts example shows, it ain’t necessarily so.
So how useful is this service for either customers or service providers?
I think it can potentially be useful but first we have to admit that the data is imperfect. It’s at an aggregate level so we can’t tell when demand peaks during a day or whether Sunday’s slower than Saturday because the museum has shorter hours. Also, it is a biased sample. This just shows when Foursquare users check in. Unless the fraction of all visitors who use Foursquare is constant over the week, the actual congestion at the museum could differ systematically from what is shown here.
Acknowledging those shortcomings, how would we expect customers to react if they were given accurate data? Suppose for the moment that all they care about is avoiding congestion (i.e., everyone wants to be alone with the Elgin Marbles) and that they don’t have inherent preferences over dates (i.e., Monday is just as convenient as any other day). Then all these variations across date should go away and the each day would have the same level of congestion. (If you need a proof of that, I actually have a paper on the topic.) Note that from the service provider’s perspective, this would be a wonderful outcome — a smooth flow arrivals that is the same from day-to-day.
Of course, that won’t happen in practice. Cutting out of work to hit the museum on Wednesday afternoon is not feasible for most people with jobs. (Remember Lee Elia: “Eighty-five percent of the world is working. The other fifteen percent come out here.” ) Then one of two things will happen. If every customer could feasibly go to any day but prefers some days to others, then the differences in congestion between days reflects the strength of preferences over days of the week. Alternatively, if some customers are restricted to a limited set of dates (i.e., weekends only) while others have flexibility, then those with flexibility spread out, lessening the load during peak times.
These are again generally good outcomes for the service provider as both alleviate the peak. Unless there are extreme economies of scale, this should reduce cost. (If there are extreme economies of scale, the provider should open only during peak times.) Note that if the peak reflects preferences (as opposed to the ability to make a given date), then the service provider should avoid making the peak more attractive. That is, the British Museum should run special programs on Tuesdays not Saturdays in order to induce more people to move off peak.
This all assumes that congestion is bad (which is the underlying assumption of When Should I Visit). However, that is not always true. If people value being part of a crowd, then providing this information will exacerbates the peaks in demand.