Why eHarmony Should Be More Like Google
A while back Fernando Ardenghi took issue with my statement “eHarmony is innovating, continuously updating the site and matching algorithm.” Fernando argues that as it updates its matching algorithm, eHarmony should recalculate compatibility between prospectives mates in its entire big database. He seems to be saying that dating sites won’t re-index their databases in order to take advantage of improved matching capabilities.
All of a sudden it hit me. Dating site compatibility assessment systems need to be more like Google.
Think about the Google web index, which is updated every so often to improve search results. When Google publishes a new index, people who are negatively affected in the organic search results tend to get upset. If you’re a company spending thousands of dollars a month in SEO fees to show up on the first page of Google’s search results and all of a sudden you’re on page six, you’re going to understandably freak out.
I like to think of eHarmony as being somewhat like Google, in that they are very secretive about their matching system, it evolves over time, and for a certain number of singles, works well. The main difference is that we know a lot more about Google’s indexing systems than we do eHarmony.
eHarmony is slowly becoming more transparent though a series of blogs and the eHarmony Labs. (although the labs blog hasn’t been updated in a month.) I for one would like to see them talk more about the matching system. How does it work, how has it evolved, what are they learning and how is that knowledge fed back into the matching system? Part of the beauty of eHarmony is that they do most of the work, but still, I’d like to know what my $50 is doing for me each month. I’m all for protecting intellectual property, but part of me thinks increased transparency into the matching process would actually be more helpful.
The marketing people at eHarmony are grinning right now, they know it’s all about ad spend. As long as more people visit and become paying members, the “better” the service becomes. If there are more members, more people will get married, regardless if the matching system improves or not.
And pretty soon, They’ll have to answer to the SEC and then it’s all about preserving shareholder value. That’s when eHarmony jumps the shark. They’re really in a difficult place right now if you think about it. Visitors are basically flat from a year ago and I assume they are spending even more money on marketing.
If they go on this acquisition spree that everyone is talking about, are they going to buy a site for the traffic, the revenue or the cross-sell, like Match promoting Chemistry (but not the other way around?)
What happens if/when eHarmony tweaks their matching system? Do people that are potential matches get disconnected? Is there a feedback loop in place where someone says, “whoa, we just lost 134,000 matches based on that last algorithm change.” Are the matches at eHarmony improving over time, staying the same, or getting worse? How can we measure improvement? I don’t want to rely on an increase in marriages per day, it’s not nuanced enough to use as a realistic measurement.
Dr. Houran and others talk a lot about the science of matching, but it really comes down to math when you’re talking about matching millions of members. What is that threshold for “connect them they are a great match” and being right on the line between a match and not? My online dating neurosis occurs when I think about the women that I’m not connecting with because my dating services are not putting them in front of me for some reason or another. Shudder. This is the stuff that drives me crazy, the potential for missed matches. My personal example on Match is when I say I want kids, and I see a million women, but if I say I’m not sure about kids, the dating pool dries up considerably. How many of those women that say they want two kids are just saying that and would be fine without kids? I meet a TON of women in that situation.
I would love to hear from someone at eHarmony who is allowed enough leeway to talk about how they run these amazingly complex matching systems, fascinating stuff.
