Orange Cone

Economics (?!?) of McBandwidth

I spend a lot of my time writing in cafes. I prefer it because it gets me away from all of the distractions of working at home or office without the isolation of working someplace completely still, such as the library or the park. It's the classic "third space" that provides for a level of a social environment without the commitment of familiarity. For me, it has also provided another benefit: no Internet. I'm easily distractable and procrastination-prone (call it "adult ADD" or "a lack of disciple" or "the modern condition." Whatever. It's the same thing that nearly everyone I know suffers from, or believe they suffer from), so having the Net around is a perfect way to kill lots of potentially productive time while learning all kinds of fabulous trivia.

However, this disconnected bliss is changing as WiFi appears in my preferred cafes and my job is becoming more dependent on being actually connected. So I'm trying to adapt my life to use them. The Coffee Beanery (what a terrible name, btw, it sounds like something between a disease and an insult) on the corner of South University and East University in Ann Arbor had free WiFi when I was here six months ago, but today although it has a functioning access point, it doesn't seem to be routing to the Net at large. I wondered why they may have chosen to cut off Net access (if not just negligence or a loose cable or something), so I started thinking about the economics of WiFi in cafes, and specifically whether McDonald's ideas is good or not.

On first blush, the McDonald's idea seemed really bad: the point of fast food is to be fast, but the point of WiFi is to get people to settle down for extended periods. It seems backward to try to piggyback an idea that extends people's stay on a business model that depends on high turnover. But McDonald's isn't dumb, so I decided to think about what this means, and also what it means for smaller places (like Maxfield's Cafe and Samovar, my preferred free-wireless cafes in San Francisco). These locations are a new kind of ISP, one that's much more dependent on its users' specific behaviors than previous ISPs, which could think in more traditional service/commodity ways.

Short of reading The Economist every week for years—I have no economics training, so I'm sure I'm making lots of freshman mistakes here, but I decided to take a cut at thinking about it.

First, I wanted to think about how to value a single sit-down seat in a cafe. I'm explicitly not looking at the counter side of the coffee/burger business, since that seems like a different product line—the people who walk in, get a thing and run out have a different agenda than the people who come to the cafe and sit down. Presumably, this later group is the one that McDonald's is hoping to build, to be more like Maxfield's or Samovar, where all the seats are full pretty much all day (I'm imagining that some VP said "hey, we have all of these great plastic booths that only get used during the lunch rush—why don't we install WiFi so that they get used ALL the time!" as justification to get the WiFi scheme funded). I'm also not thinking of the problem of pricing the connection. That's way beyond my capabilities and I'm more interested in the general outline of problem, mostly for my own understanding (for pricing info see this discussion). I'm pretty much assuming that the connection is free and the value that it provides the cafe is in greater consumption of the cafe's products as people spend more time in the cafe and return more often. This is behavior I've actually seen in mysell. Molly and I go to Maxfield's all the time and we probably spend $12 between the two of us twice a week, as opposed to the $4 we would spend—at best—if we only visited there when we wanted a drink. McDonald's is actually charging for their connectivity (and the, confusingly, different rates in each of their test markets—which is either a bake-off between the various providers participating or an attempt to react to regions differences).

I came up with this formula to model the value of the filled seats in a cafe at a given time:

value of a filled seat = dollar value of purchase/time seat is occupied – per seat overhead

This includes the assumption that a person's value to the business is not just in their single purchase, but in their presence in the cafe, as either advertisement for the cafe OR in terms of future purchases. After they buy their cup of coffee, their value does not drop to zero, but depreciates over time. However, the longer they hang out, the less it's worth.

Second, there's an additional long-term consideration which is much like that claimed by many web sites: repeat usage increases the overall value while adding little overhead. In terms of a formula, this looks like:

long term value = value of a filled seat * (# visits per person * likelihood of return visit) * # of people

Assuming that "likelihood of return visit" is proportional to "time seat is occupied" this implies that long term value is dependent on how "time seat is occupied" affects "likelihood of return visit" versus "value of a filled seat" (since it increases one while decreasing the other). But I don't think it just cancels out, as my simplistic math implies.

OK, so what does this mean? This is getting pretty muddled (but, dammit, this is my blog and it's OK to be in over my head ;-). What's supposed to happens is that WiFi use increases "# of visits per person" and "# of people". What undeniably DOES happen is that it increases "time seat is occupied" and "per seat overhead" regardless of whether the other things go up. So there are two questions that I can't answer:

• at which point does the system pass the point where the increased value of long-term
  usage overcomes the cumulative depreciation of all the time people sitting?
  
• how does this compare to the situation where sitting time isn't counted at all?
  

And one final question: regardless of all of these calculations, if McDonald's is expecting long-term usage of their facilities, if they don't invest in more comfortable furniture at the same time as adding WiFi will they still fail as people discover that no matter how the McBandwidth, the McChairs are still made of hard McPlastic?

Posted by mikek at September 4, 2003 09:42 PM