Marketing Insights and Analysis
Valentine’s Day is nearly on upon us, and several news journals and Twitters users have mentioned anectdotal observations suggesting online dating increased as the economy has decreased this past year. While I admire this optimism, something didn’t sound right. Dating is expensive, especially in the beginning when you first meet a person and are trying new places.
My theory is that as the economy sours, people decrease participation in dating services as they look to cut discretionary expenses.
How have the dating sites performed over the last year?
Compete.com shows that my theory held true into the fourth quarter of 2008, but that Match.com started experiencing growth in November. eHarmony.com has only recently stemmed their decrease in visits.
To compare this with the economy, I charted visits Match.com and eHarmony.com against the S&P 500, and adjusted for scale:

Comparison of Match.com, eHarmony.com and S&P 500
Observing the chart, it appears my theory holds true for eHarmony.com and the S&P 500. There is a correlation of .51 between the two data sets for the past 12 months. If we exclude January 2009 (and the likely pre-Valentine’s Day seasonal bump in dating traffic), that correlation increases to .849. That’s a strong enough correlation to where I believe it supports my theory. People using eHarmony seem to mimick the behavior of the S&P 500.
Match.com paints a different story. The correlation is a weak .179, and although we considerably strenghthen the relationship to .626 when we exclude January 2009, the chart indicates a multi-month upward trend prior to January. During the same period, the S&P 500 dropped. That clearly challenges my theory.
We don’t know the source of traffic to Match.com during the upward trend. They may have increased advertising or introduced new compelling features that brought customers back. A quick look at sub-domains on Quantcast showed the only measurable increase in traffic during this time was related to picture domains – not a compelling feature – which would suggest more promotion. Still, without better source information we can’t come to a conclusion.
What does this mean?
This is an interesting comparison, but it is based on only two datasets. We don’t know what either site spent to promote traffic during this time. We’re limited in competitive web information to 1 year, so we do not know if this is seasonal. We would want 24 months to adjust for seasonality.
If you are an eHarmony user and find yourself single this Saturday evening, take heart. It’s not you, it’s the economy.
And if you’re a single Match.com user this Valentine’s Day, here’s wishing the upward trending continues your way.
Happy Valentine’s Day!
You must be logged in to post a comment.
Information is essential in generating good insights, but it cannot be a crutch in making decisions. The posts here are intended to explore and are not perfect, but that's part of the point.
Doug Schumacher
February 16th, 2009 at 12:46 pm
Rob,
Intriguing post — and funny conclusion lines.
It prompted me to take a look at Search traffic as a possible indicator.
While the terms ‘recession’ and ‘economy’ both track (just to test the system), ‘dating’ appears less in sync (although I don’t have the math chops to project your correlation numbers).
But one trend I find interesting is that ‘recession’ and ‘economy’ dropped precipitously over the holidays, while ‘dating’ spiked.
http://www.google.com/insights/search/#q=recession%2Ceconomy%2Cdating&geo=US&date=today%2012-m&cmpt=q
Maybe it’s a matter of just having some down time to search, or maybe it’s hitting the holidays dateless and feeling a little depressed about it.
Either theory could also apply to a closer look at ‘dating’ searched. There’s a consistent spike every weekend, but not for ‘recession’.
http://www.google.com/insights/search/#q=dating%2Crecession&geo=US&date=today%203-m&cmpt=q
So dating may provide needed escapism, but they also seem to be doing it on their own time. Either that or, like the holidays, the weekend comes and people freak about not having a date.
Interesting possibilities. Thx for the post.
doug
Rob
February 16th, 2009 at 1:49 pm
Interesting. You’d have to look at the yearly seasonality to see if this always happened. You’d need the multi-year historicals and at a daily volume (i had neither!), but you could calculate a much better correlation that way. It wouldn’t necessarily solve for the “recession” term – that’s a focused event that is recent and hopefully shortlived – but it would shore up the rest of the model.
Good thoughts, thanks!