Match Group CEO: Digital Dating, What Makes a Good Profile and the Loneliness Epidemic

Nicolai Tangen (0:00)

You don't want to date a looser, right?

Spencer Rascoff (0:03)

Well, we believe there's someone for everyone, so I don't pass judgment. There's no such thing as a looser. That's kind of your business model, isn't it? There's no such thing as a looser. They're just somebody that hasn't met their right person yet.

Nicolai Tangen (0:13)

That sounds really good.

Hi, everyone. I'm Nicolai Tangen, the CEO of the Norwegian Sovan Wealth Fund, and today I'm joined by somebody really special, Spencer Rascoff, the CEO of Match Group. That's the company behind Tinder, Hinge, and around a dozen of other dating apps. Now, think about it for a second here. Forty percent of couples today meet online, not for friends, not at work. So this is one of the biggest shifts in human behavior that we have ever seen, and Spencer, he sits in the middle of it. A bit of an outsider, he built the property giant Zillow before taking over Match a year and a half ago, and we, in the Norwegian Sovan Wealth Fund, own one percent of Match, so we are cheering you on, Spencer.

Spencer Rascoff (1:02)

Thank you. I am glad that you are. I appreciate it, and thank you for having me.

Nicolai Tangen (1:05)

Now, how are matches actually made?

Spencer Rascoff (1:08)

We have a recommendation algorithm, the RexALGO, which tries to figure out who to show to whom. The way I like to think about it is, there are decks of cards kind of on the table. Imagine 10 decks of cards on this table with 52 cards each. The order of each deck has already been determined by us, based on certain rules associated with each user type. Like, there might be a deck that is optimized to show people in a particular order, to keep them as a payer, essentially to retain their monetization. There might be another deck for first-time user to drive engagement based on first-time users. There might be all these different goals. What we've recently done was we've done what we call queue unification, where we've combined all of these decks into one and then resorted the cards.

That's radically improved the quality of recommendations as measured by matches and chats and sparks.

Nicolai Tangen (2:04)

So what exactly has happened there?

Spencer Rascoff (2:07)

We've basically customized and personalized the order of a much larger stack based on what we know about you as compared with other people who have exhibited similar behavior to you in the past.

Nicolai Tangen (2:18)

And is that based on who I like?

Spencer Rascoff (2:20)

Yes, exactly.

Nicolai Tangen (2:21)

So if somebody likes people with red hair, they see a lot of red hair.

Spencer Rascoff (2:24)

They will, but one of the big limitations is they'll only see people with more red hair a couple hours later. So right now, our recommendations are not real time, which is a big problem. When will it be red time?

By the end of the year, which is the way other recommendation algorithms work. If you're on Spotify and you like a certain type of music or Instagram or TikTok, you can see the algorithm changing as you give it more signal, real time, that will come later this year. We think that has a lot of potential.

Nicolai Tangen (2:51)

In addition to looks and so on, what are the other things you sort according to?

Spencer Rascoff (2:55)

Their profile information. So we've made a lot of effort over the last year and a half to make profiles more rich, so additional content. And now you can leave what's called a contextual like.

So old Tinder, you would just swipe right, which basically says, I'm interested in this person. New Tinder, the profile has lots of different information, like what do they do for work? And what's their relationship with their friends like? And what do they like to do on weekends? And when you respond, you're now indicating that you're interested in that particular item that you, you know, you're saying, yes, I like that they like to go for hikes. And so now you've not only told the user, that recipient that you like, that they like to go on hikes, but you told us that you like people that like to go on hikes.

Nicolai Tangen (3:37)

What's when you, how much data do you have?

Spencer Rascoff (3:43)

Near infinite. I mean, we have billions and billions of matches.

Nicolai Tangen (3:47)

How many people do you have drilling down in this? How many scientists are there?

Spencer Rascoff (3:50)

We have this recommendations algorithm team that just eats and sleeps. This topic is probably 40 people, including some very senior people from TikTok and other companies that have really pioneered the recommendation algorithms.