Location-aware social applications

Every few months, I hear about another startup trying to get people to hook up with others nearby using their mobile phones. This has an immense boil the ocean problem. What are the chances that two compatible people might meet each other by using this service while sitting in nearby coffee shops? Let's see.

Let's assume you're single, signed up with this service, and in a coffee shop. Let's also assume the best possible location: some young, hip area of Manhattan full of coffee shops. (This is a huge assumption whose incorrectness would be enough, alone, to kill this idea. But we can temporarily assume that you only need your service to work in major dense cities full of upper-class young singles and coffee shops.)

You're probably willing to walk about 10 blocks for a casual meetup, and maybe the other person will walk the same distance to meet you. How many coffee shops are in a 20-block radius? Well, that's a lot. Given real-world constraints, I'm willing to accept a very generous maximum of 300 coffee shops.

This city is very enthusiastic about caffeine. They'd have to be, I guess, to have 300 coffee shops in such a small area. So how many people are sitting in each coffee shop? How about 20? That seems pretty high to me, but I'll be generous. That would give you approximately 6,000 people who might be available within your radius.

For random people sitting in a coffee shop, what are the chances that:

  • They have ever heard of your service and went to the website to check it out. 1 in 100? If you're lucky.
  • They signed up with your service. Maybe 1 in 10, if your website is very good.
  • They remember to notify the service of their location. This is a big one. 1 in 5, assuming some very dedicated fans.
  • They are your desired gender. Given the nature of this service, the majority of your members are probably straight males, so I'll give you 1 in 3 here.
  • They are single now. Many people might forget to deactivate their membership after they've entered a relationship. 1 in 2.
  • They are within an appropriate age range for you. Such high-tech services are likely to be used mostly by people of ages 25-40, so I'll give you the benefit of the doubt with 1 in 2.
  • You find them attractive enough. Assuming that members of this service are at least moderately attractive (another big, fatal assumption), you can probably expect, at best, 1 in 4.
  • You find them interesting and engaging in conversation. All I can say for this one: good luck. 1 in 5, and that's being extremely generous.
The probability of all of these criteria being simultaneously met, for any given person, are 1 in 1,200,000. And remember, I was being generous on most of the probabilities. Therefore, on a given night, you have a 0.5% chance that one person out of the 6,000 will be a good match. You'd have to try this approximately 200 times, in the best circumstances, before finding someone even moderately acceptable. And that's if the service is highly successful.

That's pretty damning. But that's not all.

The matchup service is likely to send you out to find other people who don't match your qualifications, because they can't measure all of them, especially whether the person is attractive or interesting. The best they could do is show you a picture, but anyone who has ever met an internet person in real life knows that anyone can come up with one picture in which they look attractive. And you're still on your own for deciding whether the person is interesting to you.

The probability that they'll meet all of the objective criteria is only 1 in 60,000, giving you a 10% chance that the service will match you up with someone well enough that you'd want to walk over and meet them, but there's only a 5% chance that they'll be a good match. So you can only expect 1 in 20 people you meet to be moderately worthwhile. Again, this is all assuming the absolute best circumstances.

That's not too bad. But when the average customer needs to meet 19 "appearance-challenged" or insanely dull (or just insane) people before they find someone moderately acceptable, how many people are going to stick with your service long enough to have any success?

It gets worse

Observant readers should notice that I left out a critical probability factor:

  • They find you attractive and interesting. Same probabilities as above: 1 in 4, and 1 in 5, respectively.

This drops the probability of a satisfactory match to 1 in 24,000,000 per person, or 0.025% of the 6,000 people available to you. You'd still be sent to prospective matches approximately 10% of the time, but you'd need to meet 300 people on average before you encountered a mutually acceptable match. Note that this isn't guaranteed to be a great match - just acceptable.

Are people loyal enough to a matchup service that they'd sit through 299 awful, awkward encounters before finding someone who they might date for a few months?

I wouldn't bet a company on it.