3 ms·
My anecdotal user-end data-science-ish story about dating apps: A few years back I was single and on Hinge a fair amount. If you used Hinge back then, you'll r
by parliament32 2y ago
My anecdotal user-end data-science-ish story about dating apps:
A few years back I was single and on Hinge a fair amount. If you used Hinge back then, you'll remember some key differences between the platform and other dating apps: 1) when you "like"'d someone, you'd have to comment on a specific part of their profile (a photo, a prompt answer, etc), 2) these likes showed up in their inbox, independent of whether they liked you or not (as in, you didn't have to like each other mutually; the other end decided whether to reply or ignore after delivery), and 3) there was limit per day, you could like/message 8 profiles per day, no more. On average, swiping through my 8 per day, I'd generally get 1-2 new replies, which turned roughly into 3-4 first dates per month.
One of the key elements is that the inbox was time-ordered: the most recent like you received was at the top. There was discussion on the Hinge subreddit about how girls would typically only click through the top few items in their inbox daily, and if you were lower down, you were doomed to drown under the mountain of new message they're getting on top. So I figured I'd solve for "what is the optimal time of day to be blasting out my likes to ensure I end up higher in the inbox?"
You can probably see where this is going: I requested a GDPR data export, which happened to have all my conversations, time-stamped. Crunching through in Python there was something in the data I didn't really expect.. a disproportionate number of first-replies (replies to my initial like/message, that is) were around the 2-3pm bucket. Not what I would've expected (don't these people work?) but fair enough, I started doing all my swiping in those hours instead of in the evening as I usually did.
And it worked. Good god did it work. I consistently started getting replies to 70-80% of my initial messages (from the ~10% before). I was drowning in conversations to the point where I wouldn't swipe at all for days for fear of yet another conversation to manage. Within a few months I ended up meeting my current girlfriend and haven't been back on since, but it was surprising how well something simple like time-of-day affected my reply rate.
- Horffupolde 2y agohttps://xkcd.com/196/ https://xkcd.com/196/
- TheJoeMan 2y agoQuite interesting! But you’re leaving us hanging, did you ask her why she was on Hinge at 2pm?
- stanford_labrat 2y agoMy guess is that time would be the first little work-break from the post-lunch session. About an hour after lunch and you take a little break pause so you swipe on Hinge to see if you got any matches.
- ugurs 2y agoYou sound like a manager.
- parliament32 2y agoI never talked to any of the girls about this. Pretty obviously "I analyzed the data from hundreds of conversations to optimize..." is not a good look. I'm pretty confident though that it's just the after-lunch doldrums and people just.. sit around swiping at work? Best guess anyway
- TheJoeMan 2y agoIt could be deeper, maybe they start thinking about their evening plans?
- HeckFeck 2y agoAny problem can be solved with enough data munging.
- dkjaudyeqooe 2y agoThat's similar to getting votes on HN, it's mostly about appearing near the top of the comments, and that has mostly to so with getting in relatively early. You also need to be reasonably good at commenting (ie, don't be ugly).
- Dwedit 2y agoOr just reply and hijack the top comment.
- kazinator 2y ago> don't these people work? They do! 2-3 p.m. is around the time people get fidgety at work and start looking at the clock, checking their phones, and such. They are no longer at lunch. Whatever busywork they had to rush through in the morning is done.
- 65 2y agoThis is the spirit of Hacker News. Now this is hacking!
- 4gotunameagain 2y agoFunny, thank you for sharing. I wonder what the relationship with girl's age is. I would imagine there are huge differences between let's say 20,25,30 and 35+. Did you happen to group by age bucket ?!
- parliament32 2y agoI did not, because I filter on pretty strict ranges ("half your age plus seven" was my lower bound, my age was the upper bound). Not enough of a spread there to get any interesting statistics.
- volkk 2y ago> ("half your age plus seven" was my lower bound, i remember reading about this rule of thumb but dang, i'm 35 and can't imagine dating someone that's 24-25 right now. no judgments here btw.
- rendaw 2y agoIt really highlights that volume of interactions is the issue that dating apps are failing to solve.