This is bullshit.
Let’s look at the math involved in this claim. For simplicity’s sake, let’s assume that each trait is binary: either you’re compatible or you’re not with a certain person. Let’s also assume that none of these are dependent on other traits, and that people have a 50% chance of being compatible on any given trait. Let’s also assume, for the sake of argument, that these guys can accurately assess each of these traits in people. Finally, let’s note that none of the traits they consider here are “has the right genitalia,” so let’s add that as a 30th (though this doesn’t account for bisexuals). That gives you roughly one in a billion odds of finding a perfect match.
Obviously no match will be perfect, and, perhaps more perniciously, eHarmony could in theory precisely quantify just how incompatible a person is for you (I have no idea if they actually do this).
But the bigger problem is that this assumes that people can be reduced to a set of psychological parameters; even if there is some kind of unquantifiable chemistry between people that it is secondary in importance to these.
I used a different dating service and it worked out pretty well. In fact, after looking at (and in one case using) other dating services that try to boil people down to a long laundry-list of multiple-choice questions, I signed up with the one that used the fewest parameters and tried to extract the most personality (within the narrow limits of the medium). Well, except for this one maybe.
But perhaps I’m full of it. Perhaps eHarmony is a good service for people who like the idea that people can be easily quantified. In that sense, they’d have a self-selecting population to work with: its users would have already answered the “what kind of dating service do you like and how do you assess people” questions and narrowed their odds of finding a good match. Perhaps this means there needs to be a matchmaking service for dating services.