The algorithms that Match Group uses just do not work (Plenty of Fish)

We’re going to look at three potential “deal breakers” on Match Group’s Plenty of Fish (PoF) app: distance, smoking, and kids. We chose these because they are all categories you can enter as options on your profile. We looked through 110 profiles that were given to us in the Recommended section of the app (at the very bottom of the home screen), without discriminating based on looks (i.e. we looked at people we wouldn’t regularly look at). Here’s the breakdown of what we found:

Firstly, distance. PoF gives you two options for who can contact you based on distance: within 75 miles or anywhere. One assumes that if you select 75 miles, you’re not interested in people beyond 75 miles, so PoF shouldn’t show you them. Unfortunately, that is not the case. In the Recommended section of their app, less than half (45%, or 49) users lived within 75 miles of our set location. It should be noted that four people did not actually give a city/town name on their profile, putting, instead, “UK”, “England”, “somewhere”, and one with just “St”. Because we didn’t know where they were, we set their distance as 678.7 miles (the most distant part of the UK from our location) for “UK”, “somewhere”, and “St”, or 460.6 miles (the most distant part of England from our location) for the user who entered “England”.

Secondly, smoking. You can tell PoF if you’re willing to date a smoker or not. I am a smoker, which is something that I selected as an option on my profile, so one assumes that PoF should only show me people who have said they will date a smoker (or, at least, haven’t said that they wouldn’t date a smoker). We had better success with people who would date smokers than people within 75 miles. Only 28% of users (31 people) in the Recommended section said that they would not date a smoker. This does leave 72% (79 users) who would. Good for us smokers, eh?

Finally, kids. I don’t want to date a woman with children, so I’ve selected “no” in the “Would you date someone with kids?” option on my profile. We found that 58% (over half) of the people who showed up in the Recommended section had children.

In total, only 16% of users (18 out of 110 people) who PoF recommended to us passed all three criteria. That leaves 84% (92 users) who failed at least one of the deal breakers, with 8% (9 users) falling fowl of all three.

From this, we can see that the algorithm that Match Group uses to serve recommended users on PoF is seriously broken. Are we being fussy with our potential matches? Probably, but that’s not the point. These are things that Match Group makes you put on your profile. It would be nice if these things were actually used in helping you find a match. Instead, they are completely ignored by the algorithm.

This is completely contrary to what OkCupid used to be. Sure, OKCupid would show you users who you were not interested in; however, it did not have options for whether you would date a smoker or a parent in your settings; this was not information about you that OkCupid had in its database and could use when showing you users. Instead, these were options you put into your search criteria (e.g. don’t show me users with children, only show me smokers, etc), and your last search criteria would be carried over to other sections of the site (e.g. recommended users). On PoF, however, these are settings that Match Group has implemented on your profile, and Match Group doesn’t even use them. On top of this, to see whether a user would date a smoker on PoF, you have to pay Match Group £19.99 a month (something we begrudgingly did for research purposes).

Another thing to note with OKCupid was that it had more flexible options for distance. You could select 5, 10, 25, 50, 75, 100, 200 miles, or anywhere (I think those were the options), and it would usually not show you users beyond those distances (an exception would be early in the morning when there weren’t many recent updates from the UK, so it would show you updates from people in the US on the home screen). On top of this, the distance from your declared location to another user’s declared location was displayed on their profile, so you didn’t have to jump on the Google machine to see how far away they were and, thus, if it was worth your time messaging them.

If Match Group wanted to find matches for its users, you would think it would mostly recommend users who fall within the criteria that a user sets on his/her profile, not just 18 out of 110. However, Match Group is not in the business of finding you matches; rather, it’s in the business of selling adds and keeping you single. It’s in Match Group’s best interest to keep you single. If you’re single, you’re likely to stay on the site and be fed ads or become so desperate that you buy a subscription, or, god forbid, upgrade to Match.com. Whichever way you look at it, Match Group profits from your singledom; it doesn’t make money when you find a match and suspend your account. As a result, Match Group has rigged its systems to make it harder for you to find love.

2 thoughts on “The algorithms that Match Group uses just do not work (Plenty of Fish)

  1. Pingback: Strict Mode | We Miss OkCupid

  2. Pingback: We Omit OkCupid – TOP HACKER™

Leave a Reply

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

This site uses Akismet to reduce spam. Learn how your comment data is processed.