Book a call
Journal · The research

What the data actually says about dating profiles

The Match Lab · London · 9 min read

Everyone has an opinion about dating profiles. Far fewer have read the data. So we did — OkCupid's famous numbers, Photofeeler's photo studies, Hinge's own reports and a stack of academic papers. The honest summary: your first photo gets the swipe, your words get the date, and being accurate protects both. Here's the evidence, decoded.

01. Your first photo does almost all of the early work

It feels shallow because it is. On a swipe app, the first image is the whole audition.

~20% vs ~2% A 2025 dating-app study (445 users, 5,340 swipe decisions) found a meaningful bump in photo attractiveness produced roughly a 20% increase in matches — while a comparable improvement to the bio moved matches only ~2%. 2025 dating-app field study

And people decide fast. Photofeeler's research suggests strangers form an impression of your attractiveness, competence and trustworthiness in under two seconds — and more than half of swipers won't even look at the rest of your profile if the first photo doesn't land. The takeaway isn't "be hotter." It's: make your first photo legible and warm, because it's carrying the whole profile on its back.

02. But "attractive" isn't a beauty contest

Here's where the popular advice gets it wrong. OkCupid famously claimed you should avoid smiling and eye contact. Photofeeler re-ran the analysis on a far larger dataset and found the opposite-ish: smiling vs not made no significant difference — except that eye contact with no smile actually hurt, and anything covering your face (yes, sunglasses) consistently lowered scores, because we read the eyes to build connection.

The variance effect OkCupid's data showed that polarising looks often beat conventionally "pretty." A face that some people rate a 5 and others rate a 1 tends to get more messages than one everyone agrees is a safe 4 — because the people who love it message hard. Being a bit divisive beats being forgettable. OkCupid / "Dataclysm," Christian Rudder

So the goal of a good shoot isn't a flawless, universally-acceptable headshot. It's to amplify the specific thing that makes you magnetic — and let it be a little distinctive.

03. Looks win the swipe — not the connection

The most freeing stat in all of this: when OkCupid temporarily hid every profile photo, people who went on resulting blind dates reported roughly the same satisfaction with each other regardless of how "attractive" they'd been rated. Looks determine who you match with. They barely predict who you'll actually like.

Which is the whole argument for fixing your profile rather than fixing your face: better photos don't change you — they get the right person past the swipe and into the conversation where the real you can do its thing.

04. Photos get the match. Words get the date.

Attraction opens the door; specificity walks you through it. A 2022 study in PLOS ONE found that original profile text made people seem more intelligent and funny — which in turn made them more attractive and raised others' intention to date them. Clichés ("love to laugh, partner in crime, good vibes") do the opposite: they make you generic.

Hinge's own data is the clincher on conversion:

2× more dates On Hinge, Likes that come with a comment are twice as likely to lead to a date — and 72% of daters say they're more likely to consider someone who comments on something specific. Likes on text prompts were reported ~47% more likely to lead to a date than likes on photos. Hinge
+62% likes Hinge reports profiles with a video prompt get 62% more likes and 3.4× more conversation-starters; and 65% of daters say hearing a voice helps them gauge interest in a match. Hinge newsroom

Translation: your profile should hand people specific, easy things to react to. A bio that's just a vibe gets a like; a bio with a hook gets a message.

05. Make it about them — and keep it honest

Researchers at Berkeley's Haas School found most profiles are accidental monologues: over half of writers signalled "get to know me," while only about a fifth signalled wanting to know the other person. Profiles that show genuine curiosity about a partner read as warmer and less self-absorbed — and perform better.

And resist the urge to over-edit, because the apps are not the date:

54% & 62% In one survey, 54% of Gen Z and 62% of millennials said they'd ended a date early or refused a second one because the person looked significantly different from their photos. A misleading photo gets the match and kills the date. Hily survey

This is our whole rule, in one stat: flattering, not fake. The best photo is the one a real-life date confirms — "yes, that's them, on a great day" — not "wait, you look totally different."

06. You can't see yourself clearly (nobody can)

People are surprisingly bad at choosing their own best photos — you're too used to your own face, and you over-rate the picture you like. Studies show strangers do a better job selecting images that balance attractiveness, competence and trustworthiness — the exact trio Photofeeler's million-image model (Photofeeler-D3) was built to predict. The same face, in a different frame, reads as a different person.

It also matters because of who you're actually swiping on. Bruch & Newman's analysis of dating behaviour found people pursue partners about 25% more desirable than themselves — and your reply odds fall as that gap widens. Better presentation is how you compete in the pool you actually want.

Put it together and the line-up writes itself: 4–6 photos (fewer reads as low-effort; more brings decision fatigue), led by one clear, warm hero shot, then variety — full-length, "in your element," a touch of social proof — and words that give people something specific to say. That's not luck. It's design.

Want your profile built on this, not vibes?

We're a London fashion photographer + a dating-app insider. We shoot the photos, write the words, and order it all to convert.

Book a free intro call →

Sources

OkCupid data & the "variance" and blind-date findings: Christian Rudder, Dataclysm (NPR summary). · Smiling, eye contact & sunglasses: Photofeeler and their "perfect photo" study. · Photo-impression modelling: Photofeeler-D3 (arXiv). · Prompts, comments, video & voice: Hinge newsroom & Hinge prompt data. · Original-text effect: 2022 PLOS ONE study. · "Know me vs know you": Berkeley Haas research. · Photo-accuracy: Hily survey. · Aspirational matching: Bruch & Newman, Science Advances (2018). · Photo count: aggregated Tinder/Hinge/photo-tool guidance.

Some figures are reported by the platforms themselves or summarised in secondary sources; we've attributed each so you can check it.