The Verge’s OkCupid interview

The Verge is kicking off a series interviewing startup companies called Small Empires, and their most recent interview was with the rather large empire of OkCupid. The interview is fascinating to me (of course), but I think it might also fascinate singles who use OkCupid and its competitor online dating services. There’s a lot to unpack here; I’m coming from the perspective of someone who’s been obsessively studying online dating for a decade now, both for my personal life and for my business as an online dating coach. Go watch the video, then come back so we can pick it apart ruthlessly, haha.

Around minute four, Christian Rudder talks about those maddening, infuriating match percentage numbers. He expressly states that the goal of OkCupid’s match percentages is to get you out on an initial coffee date. But how many of you have relied on that match percentage as an indicator of longer-term, bigger-picture compatibility? How many of you have come to me surprised that the person with whom you shared 96% alleged simpatico levels turned out to be a douche? How many of you have ruled out going on that coffee date at all because someone’s percentage wasn’t as high as other potential dates? I’ve been saying for YEARS to ignore the match percentages in terms of deciding how worthy of your time and attention a person actually is in person, and how likely you actually are to get along in real life. But users of the site implicitly read MUCH more into that data set than the founders specifically state is intended. I have to wonder if they would change the way they displayed or discussed such features if they employed the help of a user experience guru, you know? Someone who could help them see that their users trust the sometimes fuzzy data with which they’re presented.

Around minute nine, Christian also expresses a concern that there’s a big human-sized hole in most “big data,” such as Twitter. But he claims OkCupid data is more solid because people have opted in with details like gender and height and so forth. This is a sound point; Twitter certainly doesn’t learn as much about its users during onboarding. But do you really think OkCupid’s data doesn’t have any human-shaped holes? People regularly write about themselves in ways that aren’t fully accurate, and they assign themselves metadata characteristics (diet, height, body type, income, even industry) that don’t match up with the real world. (For a while there, OkCupid actually listed peoples’ height as an inch shorter than they had stated, because that was the average data point by which they found this metric to be misreported.) It may be true that users are happy to disclose more honestly when answering Match questions than with, say, a marketing survey, but the sheer number of people concerned about online dating lies and scams should warn us that user-harvested data isn’t exactly infallible.  I agree that Big Data has holes, but I wouldn’t exempt any online dating company from that statement; the holes are just different ones, ya know?

Later in the interview, Christian states that much of the data they collect is non-actionable, and almost immediately afterwards T-shirt states that every single data point is meaningful, haha. Look, I get it. You collect a lot of info about people and can infer a ton based on that. And I’m quite certain there are some semi-nefarious marketing company deals that get to harvest that data to better understand demographics. (That’s not the reason behind my advice to use a separate email address for your online dating accounts, but if you’re the privacy paranoid type, it sure as hell can’t hurt!) Let me be clear, too, that I have a much greater respect for data-based design and decision making than I used to. Working at Amazon taught me a LOT about data and analysis and quantifying and nailing things down. But at the same time, I feel like a huge data set that’s being ignored is all the counterintuitive ways in which people misinterpret and misuse data on a site like this, and all the many unknown X-factors that no amount of data harvesting can ever really answer. After all, Christian quips that once you’re on said coffee date, there’s all this other stuff such as “chemistry” that factors in. To put it bluntly, I believe users assume that match percentage correlates directly with likelihood for in-person chemistry. That is the opposite of what the founder stated here. There’s a fundamental disconnect going on, and I wish they’d spend a little more time addressing bigger-picture design and direction issues like that.

At one point T-shirt says almost boastfully that the site has no QA. AAAAGH !@$% this makes me insane, haha. Not just as a former tester, but also as a coach and a former user. Plenty of users pay a premium (known as A-List) for features such as anonymous browsing, but OkCupid has repeatedly rolled out features that broke this alleged anonymity. (Do you think they went through and apologized or offered refunds? Ha!) They regularly tinker with and change things that drastically impact the community, yet they avoid changing features (like insecure logins) that might benefit from some focused testing. And let’s keep in mind the notoriously terrible UI design. Testing isn’t just about functionality, heh. Some user experience testing would go a long way on this site, but it’s clear that’s not as much their priority. (Heck, the shot at the end of this video showing “casual sex” is actually outdated; the euphemism in place now is “activity partners.”)

Even the way they designed their initial algorithms seems untested—and not nearly diverse enough. I mean, look at the shot of their development team. Look at the team site. Not exactly balanced. I recognize that diversity is a problem in the tech world in general, but that’s a hell of a lot of white dudes with math degrees, ya know? (And I believe their team was even more white-straight-math-guy when it was initially founded.) If they had consulted with a broader range of users, designers, developers, etc. before hacking it all together with their own paths to dating in mind, they may have come up with a different set of answers. (You guys have no idea how many women are offended and/or freaked out by the tenor of most of the Match questions. I bet we could make a great drinking game out of guessing which ones were penned by the initial white guy founders, and which ones were added later by an expanded user base.) I’m not a diversity expert, and it isn’t my usual hot-button issue in tech OR online dating, but it feels like a glaring, shall we say, human-sized hole in their approach.

Oh hey, brief industry gossip time! Did any of y’all catch that total jab at competitor HowAboutWe around minute nineteen? They basically said that users exiting the site because they’d found love was not a real problem. HowAboutWe has been (foolishly, IMO) candid about how they saw this as a major monetization issue so they basically built in a couples section to keep people in their paid system. Whether this is smarmy or not, it comes off as such—OkCupid’s delight in users opting out speaks more positively to their core values, at least as far as interview PR goes. But, BUT! ARGH! Did you HEAR him describe how painfully awkwardly awful it is for users to opt out and actually inform them that they’ve met someone? WHY NOT HIRE A USER EXPERIENCE PRO TO MAKE IT LESS SHITTY FOR YOUR USERS? Ahem. I don’t mean to get all ragecapsy. But, like, how are you this intently intelligent and aware of human behavior on the one hand, and so utterly blasé about the aspects of your site you admit totally suck for everyone? My inner perfectionist absolutely cannot reconcile this. Don’t deliberately make things harder for people. Aaaagh.

Throughout the whole interview, it struck me how Christian Rudder sort of down-talked his own customer base. He was kind of flippant about SparkNotes users and quiz-takers from back in the day, and he seemed derisive to me discussing inherently racially biased patterns in OkCupid users. (I can understand derision about openly racist behavior, but implicit biases are so ingrained in our lives, and uh, our HIRING PRACTICES, BUDDY.) I might be reading too much into it, but the language and tone he used seemed to convey a sort of disdain for the singles who made his company successful. Near the end, in talking about how OkCupid grows its user base, he states that for couples who meet online and get married, “every fool at their wedding” knows it was from their site. Every fool? Really? Like, now we’re just being derisive about guests at a wedding, you know, to celebrate the joyous union your site helped arrange? Maybe that’s just how he talks. 🙂

Christian also regularly talked about how great it was to be able to offer a better product, “for free.” Um, guys, OkCupid definitely still has a free tier, but it’s been offering paid levels for ages now. I’m certain IAC backing is part of why they now push those paid features all the dang time. So while it’s free, it’s seemingly better when paid (arguable, but that’s what their many marketing emails and UI cues would have you believe). I’m a little disappointed that we didn’t hear them address any of the potential conflicts of values or direction that happen when your startup gets bought out by a well-funded behemoth who wants to start monetizing your previously free thing. Isn’t that kind of a major element of startup trajectory? I don’t mean to criticize the Verge’s interviewing, I’m just surprised they didn’t want to address that more. Seems like the Match buyout would be the aspect that Verge readers would find most interesting.

Lastly, let’s end on a positive—props to The Verge for featuring a gay couple as their interview subjects. Even if OkCupid isn’t as great as they could be when it comes to diversity and implicit impressions on users and all that jazz, it’s awesome to see a slightly less normative perspective in the overall piece. (Now next time, send a woman to ask slightly different questions in the interview, haha. Oooh, ooh, pick me!) [icon-heart]

Leave a Reply

Your email address will not be published. Required fields are marked *