17:28
TEDSalon NY2013

Amy Webb: How I hacked online dating

Filmed:

Amy Webb was having no luck with online dating. The dates she liked didn't write her back, and her own profile attracted crickets (and worse). So, as any fan of data would do: she started making a spreadsheet. Hear the story of how she went on to hack her online dating life -- with frustrating, funny and life-changing results.

- Digital strategist
Amy Webb heads the digital strategy house Webbmedia Group, and is a founder of the SparkCamp discussion series. She's the author of "Data: A Love Story." Full bio

So my name is Amy Webb,
00:12
and a few years ago I found myself at the end
00:15
of yet another fantastic relationship
00:17
that came burning down in a spectacular fashion.
00:20
And I thought, you know, what's wrong with me?
00:23
I don't understand why this keeps happening.
00:26
So I asked everybody in my life
00:28
what they thought.
00:30
I turned to my grandmother,
00:32
who always had plenty of advice,
00:33
and she said, "Stop being so picky.
00:36
You've got to date around.
00:39
And most importantly,
00:40
true love will find you when you least expect it."
00:42
Now as it turns out,
00:46
I'm somebody who thinks a lot about data,
00:48
as you'll soon find.
00:50
I am constantly swimming in numbers
00:52
and formulas and charts.
00:55
I also have a very tight-knit family,
00:56
and I'm very, very close with my sister,
00:59
and as a result, I wanted to have
01:01
the same type of family when I grew up.
01:02
So I'm at the end of this bad breakup,
01:04
I'm 30 years old,
01:06
I figure I'm probably going to have
01:08
to date somebody for about six months
01:09
before I'm ready to get monogamous
01:11
and before we can sort of cohabitate,
01:13
and we have to have that happen for a while before we can get engaged.
01:15
And if I want to start having children by the time I'm 35,
01:18
that meant that I would have had to have been
01:21
on my way to marriage five years ago.
01:23
So that wasn't going to work.
01:26
If my strategy was to least-expect my way
01:27
into true love, then the variable that I had
01:30
to deal with was serendipity.
01:32
In short, I was trying to figure out, well,
01:34
what's the probability of my finding Mr. Right?
01:36
Well, at the time I was living in the city of Philadelphia,
01:40
and it's a big city, and I figured,
01:42
in this entire place, there are lots of possibilities.
01:45
So again, I started doing some math.
01:48
Population of Philadelphia: It has 1.5 million people.
01:50
I figure about half of that are men,
01:54
so that takes the number down to 750,000.
01:55
I'm looking for a guy between the ages of 30 and 36,
01:58
which was only four percent of the population,
02:01
so now I'm dealing with the possibility of 30,000 men.
02:04
I was looking for somebody who was Jewish,
02:07
because that's what I am and that was important to me.
02:08
That's only 2.3 percent of the population.
02:11
I figure I'm attracted to maybe one out of 10
02:13
of those men,
02:15
and there was no way I was going
02:18
to deal with somebody who was an avid golfer.
02:20
So that basically meant there were 35 men for me
02:22
that I could possibly date
02:25
in the entire city of Philadelphia.
02:28
In the meantime, my very large Jewish family
02:32
was already all married and well on their way
02:35
to having lots and lots of children,
02:38
and I felt like I was under tremendous peer pressure
02:40
to get my life going already.
02:42
So if I have two possible strategies at this point
02:44
I'm sort of figuring out.
02:46
One, I can take my grandmother's advice
02:47
and sort of least-expect my way
02:50
into maybe bumping into the one
02:52
out of 35 possible men in the entire
02:55
1.5 million-person city of Philadelphia,
02:57
or I could try online dating.
03:01
Now, I like the idea of online dating,
03:04
because it's predicated on an algorithm,
03:06
and that's really just a simple way of saying
03:08
I've got a problem, I'm going to use some data,
03:10
run it through a system
03:12
and get to a solution.
03:13
So online dating is the second most popular way
03:16
that people now meet each other,
03:18
but as it turns out, algorithms have been around
03:20
for thousands of years in almost every culture.
03:22
In fact, in Judaism, there were matchmakers
03:25
a long time ago, and though
03:28
they didn't have an explicit algorithm per se,
03:29
they definitely were running through formulas in their heads,
03:32
like, is the girl going to like the boy?
03:34
Are the families going to get along?
03:36
What's the rabbi going to say?
03:38
Are they going to start having children right away?
03:39
And the matchmaker would sort of think through all of this,
03:42
put two people together, and that would be the end of it.
03:45
So in my case, I thought,
03:47
well, will data and an algorithm
03:49
lead me to my Prince Charming?
03:52
So I decided to sign on.
03:54
Now, there was one small catch.
03:56
As I'm signing on to the various dating websites,
03:57
as it happens, I was really, really busy.
04:00
But that actually wasn't the biggest problem.
04:02
The biggest problem is that I hate
04:05
filling out questionnaires of any kind,
04:07
and I certainly don't like questionnaires
04:09
that are like Cosmo quizzes.
04:11
So I just copied and pasted from my résumé.
04:13
(Laughter)
04:16
So in the descriptive part up top,
04:21
I said that I was an award-winning journalist
04:24
and a future thinker.
04:26
When I was asked about fun activities and
04:28
my ideal date, I said monetization
04:30
and fluency in Japanese.
04:33
I talked a lot about JavaScript.
04:36
So obviously this was not the best way
04:38
to put my most sexy foot forward.
04:42
But the real failure was that
04:45
there were plenty of men for me to date.
04:47
These algorithms had a sea full of men
04:50
that wanted to take me out on lots of dates --
04:52
what turned out to be truly awful dates.
04:55
There was this guy Steve, the I.T. guy.
04:59
The algorithm matched us up
05:01
because we share a love of gadgets,
05:03
we share a love of math and data and '80s music,
05:05
and so I agreed to go out with him.
05:09
So Steve the I.T. guy invited me out
05:11
to one of Philadelphia's white-table-cloth,
05:13
extremely expensive restaurants.
05:15
And we went in, and right off the bat,
05:17
our conversation really wasn't taking flight,
05:19
but he was ordering a lot of food.
05:22
In fact, he didn't even bother looking at the menu.
05:24
He was ordering multiple appetizers,
05:26
multiple entrées, for me as well,
05:28
and suddenly there are piles and piles of food on our table,
05:31
also lots and lots of bottles of wine.
05:33
So we're nearing the end of our conversation
05:36
and the end of dinner, and I've decided
05:38
Steve the I.T. guy and I are really just not meant for each other,
05:39
but we'll part ways as friends,
05:42
when he gets up to go to the bathroom,
05:44
and in the meantime the bill comes to our table.
05:47
And listen, I'm a modern woman.
05:50
I am totally down with splitting the bill.
05:53
But then Steve the I.T. guy didn't come back. (Gasping)
05:56
And that was my entire month's rent.
06:00
So needless to say, I was not having a good night.
06:04
So I run home, I call my mother, I call my sister,
06:08
and as I do, at the end of each one of these
06:11
terrible, terrible dates,
06:14
I regale them with the details.
06:16
And they say to me,
06:18
"Stop complaining."
06:20
(Laughter)
06:22
"You're just being too picky."
06:23
So I said, fine, from here on out
06:26
I'm only going on dates where I know
06:28
that there's wi-fi, and I'm bringing my laptop.
06:29
I'm going to shove it into my bag,
06:31
and I'm going to have this email template,
06:33
and I'm going to fill it out and collect information
06:35
on all these different data points during the date
06:37
to prove to everybody that empirically,
06:40
these dates really are terrible. (Laughter)
06:42
So I started tracking things like
06:44
really stupid, awkward, sexual remarks;
06:46
bad vocabulary;
06:49
the number of times a man forced me to high-five him.
06:51
(Laughter)
06:54
So I started to crunch some numbers,
06:56
and that allowed me to make some correlations.
07:00
So as it turns out,
07:03
for some reason, men who drink Scotch
07:06
reference kinky sex immediately.
07:09
(Laughter)
07:11
Well, it turns out that these
07:13
probably weren't bad guys.
07:16
There were just bad for me.
07:17
And as it happens, the algorithms that were setting us up,
07:19
they weren't bad either.
07:23
These algorithms were doing exactly
07:24
what they were designed to do,
07:26
which was to take our user-generated information,
07:27
in my case, my résumé,
07:30
and match it up with other people's information.
07:32
See, the real problem here is that,
07:35
while the algorithms work just fine,
07:36
you and I don't, when confronted
07:38
with blank windows where we're supposed
07:40
to input our information online.
07:42
Very few of us have the ability
07:44
to be totally and brutally honest with ourselves.
07:46
The other problem is that these websites are asking us
07:50
questions like, are you a dog person or a cat person?
07:52
Do you like horror films or romance films?
07:55
I'm not looking for a pen pal.
07:58
I'm looking for a husband. Right?
08:00
So there's a certain amount of superficiality in that data.
08:02
So I said fine, I've got a new plan.
08:05
I'm going to keep using these online dating sites,
08:08
but I'm going to treat them as databases,
08:10
and rather than waiting for an algorithm to set me up,
08:12
I think I'm going to try reverse-engineering this entire system.
08:15
So knowing that there was superficial data
08:19
that was being used to match me up with other people,
08:22
I decided instead to ask my own questions.
08:24
What was every single possible thing
08:27
that I could think of that I was looking for in a mate?
08:28
So I started writing and writing and writing,
08:31
and at the end, I had amassed
08:36
72 different data points.
08:38
I wanted somebody was Jew...ish,
08:41
so I was looking for somebody who had the same
08:43
background and thoughts on our culture,
08:45
but wasn't going to force me to go to shul
08:47
every Friday and Saturday.
08:49
I wanted somebody who worked hard,
08:51
because work for me is extremely important,
08:53
but not too hard.
08:55
For me, the hobbies that I have
08:56
are really just new work projects that I've launched.
08:58
I also wanted somebody who not only wanted two children,
09:01
but was going to have the same attitude toward parenting that I do,
09:04
so somebody who was going to be totally okay
09:07
with forcing our child to start taking piano lessons at age three,
09:09
and also maybe computer science classes
09:12
if we could wrangle it.
09:16
So things like that, but I also wanted somebody
09:18
who would go to far-flung, exotic places,
09:20
like Petra, Jordan.
09:22
I also wanted somebody who would weigh
09:23
20 pounds more than me at all times,
09:25
regardless of what I weighed.
09:27
(Laughter)
09:28
So I now have these 72 different data points,
09:31
which, to be fair, is a lot.
09:34
So what I did was, I went through
09:36
and I prioritized that list.
09:37
I broke it into a top tier and a second tier of points,
09:39
and I ranked everything starting at 100
09:43
and going all the way down to 91,
09:46
and listing things like I was looking for somebody who was really smart,
09:48
who would challenge and stimulate me,
09:51
and balancing that with a second tier
09:53
and a second set of points.
09:55
These things were also important to me
09:57
but not necessarily deal-breakers.
09:59
So once I had all this done,
10:04
I then built a scoring system,
10:05
because what I wanted to do
10:07
was to sort of mathematically calculate
10:09
whether or not I thought the guy that I found online
10:11
would be a match with me.
10:14
I figured there would be a minimum of 700 points
10:15
before I would agree to email somebody
10:18
or respond to an email message.
10:19
For 900 points, I'd agree to go out on a date,
10:21
and I wouldn't even consider any kind of relationship
10:23
before somebody had crossed the 1,500 point threshold.
10:26
Well, as it turns out, this worked pretty well.
10:30
So I go back online now.
10:33
I found Jewishdoc57
10:35
who's incredibly good-looking, incredibly well-spoken,
10:38
he had hiked Mt. Fuji,
10:41
he had walked along the Great Wall.
10:42
He likes to travel as long as it doesn't involve a cruise ship.
10:44
And I thought, I've done it!
10:47
I've cracked the code.
10:50
I have just found the Jewish Prince Charming
10:52
of my family's dreams.
10:56
There was only one problem:
10:58
He didn't like me back.
11:00
And I guess the one variable that I haven't considered
11:03
is the competition.
11:06
Who are all of the other women
11:08
on these dating sites?
11:10
I found SmileyGirl1978.
11:12
She said she was a "fun girl who is Happy and Outgoing."
11:16
She listed her job as teacher.
11:19
She said she is "silly, nice and friendly."
11:20
She likes to make people laugh "alot."
11:23
At this moment I knew, clicking after profile
11:25
after profile after profile that looked like this,
11:28
that I needed to do some market research.
11:30
So I created 10 fake male profiles.
11:32
Now, before I lose all of you --
11:35
(Laughter) --
11:38
understand that I did this
11:40
strictly to gather data
11:44
about everybody else in the system.
11:46
I didn't carry on crazy Catfish-style relationships with anybody.
11:47
I really was just scraping their data.
11:52
But I didn't want everybody's data.
11:54
I only wanted data on the women
11:56
who were going to be attracted
11:58
to the type of man that I really, really wanted to marry. (Laughter)
11:59
When I released these men into the wild,
12:02
I did follow some rules.
12:06
So I didn't reach out to any woman first.
12:07
I just waited to see who these profiles were going to attract,
12:09
and mainly what I was looking at was two different data sets.
12:12
So I was looking at qualitative data,
12:16
so what was the humor, the tone,
12:17
the voice, the communication style
12:19
that these women shared in common?
12:21
And also quantitative data,
12:23
so what was the average length of their profile,
12:24
how much time was spent between messages?
12:26
What I was trying to get at here was
12:28
that I figured in person,
12:30
I would be just as competitive
12:32
as a SmileyGirl1978.
12:33
I wanted to figure out how to maximize
12:35
my own profile online.
12:37
Well, one month later,
12:40
I had a lot of data, and I was able to do another analysis.
12:42
And as it turns out, content matters a lot.
12:46
So smart people tend to write a lot --
12:49
3,000, 4,000,
12:51
5,000 words about themselves,
12:53
which may all be very, very interesting.
12:55
The challenge here, though, is that
12:57
the popular men and women
12:58
are sticking to 97 words on average
13:00
that are written very, very well,
13:03
even though it may not seem like it all the time.
13:05
The other sort of hallmark of the people who do this well
13:08
is that they're using non-specific language.
13:10
So in my case, you know,
13:12
"The English Patient" is my most favorite movie ever,
13:14
but it doesn't work to use that in a profile,
13:17
because that's a superficial data point,
13:21
and somebody may disagree with me
13:22
and decide they don't want to go out with me
13:24
because they didn't like sitting through the three-hour movie.
13:25
Also, optimistic language matters a lot.
13:28
So this is a word cloud
13:30
highlighting the most popular words that were used
13:32
by the most popular women,
13:34
words like "fun" and "girl" and "love."
13:36
And what I realized was not that I had
13:38
to dumb down my own profile.
13:40
Remember, I'm somebody who said
13:42
that I speak fluent Japanese and I know JavaScript
13:43
and I was okay with that.
13:46
The difference is that it's about being more approachable
13:48
and helping people understand
13:51
the best way to reach out to you.
13:53
And as it turns out, timing is also really, really important.
13:55
Just because you have access
13:57
to somebody's mobile phone number
13:59
or their instant message account
14:01
and it's 2 o'clock in the morning and you happen to be awake,
14:03
doesn't mean that that's a good time to communicate with those people.
14:05
The popular women on these online sites
14:08
spend an average of 23 hours
14:11
in between each communication.
14:13
And that's what we would normally do
14:15
in the usual process of courtship.
14:17
And finally, there were the photos.
14:19
All of the women who were popular
14:22
showed some skin.
14:24
They all looked really great,
14:26
which turned out to be in sharp contrast
14:28
to what I had uploaded.
14:31
Once I had all of this information,
14:34
I was able to create a super profile,
14:36
so it was still me,
14:38
but it was me optimized now for this ecosystem.
14:40
And as it turns out, I did a really good job.
14:44
I was the most popular person online.
14:49
(Laughter)
14:52
(Applause)
14:54
And as it turns out, lots and lots of men wanted to date me.
14:58
So I call my mom, I call my sister, I call my grandmother.
15:02
I'm telling them about this fabulous news,
15:04
and they say, "This is wonderful!
15:06
How soon are you going out?"
15:08
And I said, "Well, actually, I'm not going to go out with anybody."
15:10
Because remember, in my scoring system,
15:13
they have to reach a minimum threshold of 700 points,
15:15
and none of them have done that.
15:17
They said, "What? You're still being too damn picky."
15:19
Well, not too long after that,
15:22
I found this guy, Thevenin,
15:24
and he said that he was culturally Jewish,
15:26
he said that his job was an arctic baby seal hunter,
15:28
which I thought was very clever.
15:31
He talked in detail about travel.
15:34
He made a lot of really interesting cultural references.
15:37
He looked and talked exactly like what I wanted,
15:39
and immediately, he scored 850 points.
15:42
It was enough for a date.
15:45
Three weeks later, we met up in person
15:47
for what turned out to be a 14-hour-long conversation
15:49
that went from coffee shop to restaurant
15:52
to another coffee shop to another restaurant,
15:55
and when he dropped me back off at my house that night
15:57
I re-scored him --
15:59
[1,050 points!] --
16:00
thought, you know what,
16:02
this entire time I haven't been picky enough.
16:03
Well, a year and a half after that,
16:06
we were non-cruise ship traveling
16:08
through Petra, Jordan,
16:11
when he got down on his knee and proposed.
16:13
A year after that, we were married,
16:16
and about a year and a half after that, our daughter,
16:19
Petra, was born.
16:21
(Applause)
16:23
Obviously, I'm having a fabulous life, so --
16:30
(Laughter) --
16:33
the question is, what does all of this mean for you?
16:34
Well, as it turns out, there is an algorithm for love.
16:37
It's just not the ones that we're being presented with online.
16:40
In fact, it's something that you write yourself.
16:44
So whether you're looking for a husband or a wife
16:46
or you're trying to find your passion
16:49
or you're trying to start a business,
16:51
all you have to really do is figure out your own framework
16:52
and play by your own rules,
16:55
and feel free to be as picky as you want.
16:57
Well, on my wedding day,
17:00
I had a conversation again with my grandmother,
17:01
and she said, "All right, maybe I was wrong.
17:03
It looks like you did come up with
17:06
a really, really great system.
17:07
Now, your matzoh balls.
17:09
They should be fluffy, not hard."
17:12
And I'll take her advice on that.
17:15
(Applause)
17:17

▲Back to top

About the Speaker:

Amy Webb - Digital strategist
Amy Webb heads the digital strategy house Webbmedia Group, and is a founder of the SparkCamp discussion series. She's the author of "Data: A Love Story."

Why you should listen

Amy Webb tells stories with data -- in her award-winning practice as a reporter for Newsweek and the Wall Street Journal, and now as the head of Webbmedia Group, a digital-strategy consultancy.

She's a co-founder of SparkCamp, a series of weekend discussions about big ideas around media: from great new storytelling styles to making world change.

Her book Data: A Love Story tells the tale of how she gamed the online dating system to figure out how to find the love of her life. 

More profile about the speaker
Amy Webb | Speaker | TED.com