Beyond the Algorithm: How Psychological Affinity Matching Actually Works

Product · by Simone Rainieri · 8 min read

Most matching algorithms in 2026 are sorting hats with a marketing budget.

"You both like coffee and travel." Great. So does most of the planet. That is a census statistic, not a match.

Last Thursday, 6pm, Costa Coffee in South Kensington. Lukewarm espresso in one hand, phone in the other. Two women next to me were comparing dating-app matches. "He likes dogs, hiking, and The Office." "Which series?" "Does it matter? They all say the same thing."

She is right. It does not matter. Interest-based matching tends to produce drift: a loop of polite small talk that goes nowhere.

This piece explains how Vairi's psychological affinity matching works, why we match on conflict style rather than hobbies, and what the research actually says about what keeps people close.

Why interest-based matching falls short

Interest-based matching pairs people on stated preferences: hobbies, the shows they watch, lifestyle. It feels obvious, but shared interests are a weak predictor of how deep a relationship gets. Plenty of people feel lonely in a room full of others who like the same films.

The research on relationship formation keeps landing in the same place: shared interests predict who starts talking, not who stays close. They are a surface signal. Useful for the first message, not much help after that.

What does predict quality is duller and harder to fake: how people communicate, how they handle conflict, how they manage their own emotions. That is the engine. The rest is paint.

Insomma, it is less about what you like and more about how you process the world. That is the gap between a real bond and a quiet evening of drift.

Shared interests start conversations; shared psychology is what sustains them. Connection needs some structural alignment, not just similar shopping habits.

The three dimensions of affinity

Vairi matches on three psychological dimensions, picked up through a conversational onboarding rather than a quiz. Quizzes are fastidiosissimo: clinical and cold.

Conflict style is how you handle disagreement. Are you a confronter who names the issue straight away, an avoider who protects the harmony, or a processor who needs time to think first? We do not pair identical styles, we pair compatible ones. A confronter with a processor tends to work. Two confronters mostly produces friction.

Energy orientation is how you recharge socially, and it is more than introvert versus extrovert. We look at how many deep connections you can sustain at once, and how much recovery time you need between intense conversations before you are wiped out.

Temporal mode is whether you live mostly in the past and reflect, in the present and experience, or in the future and plan. Two present-oriented people have a brilliant time in the moment but can struggle to build continuity. A past-and-future pairing creates a natural pull that keeps a conversation going for weeks.

It is the same idea as semantic search: matching on meaning instead of surface words. We do that with people, looking for the intent underneath the keywords of your life.

The onboarding

Beh, I will be honest: our onboarding is not short. About 5 to 7 minutes. We ask your birthday and what you are here for, then run a behavioural read disguised as a chat with Pidge, which sidesteps the form fatigue most social apps drown in.

Last Wednesday a user called Tom told me about it. He is 29, lives in Clapham, an accountant who always has a battered Kindle on him. He said: "I nearly quit the onboarding because it felt long. Then I realised it was the first app that asked how I think, not just what I like. That felt respectful."

The onboarding feeds a profile that pairs you with someone who will both challenge you and put you at ease. We are not looking for your twin. We are looking for your complement, which is what keeps a match from sliding into drift.

A good match is not someone who agrees with everything you say. It is someone who makes you think something you had not thought before. Friction helps, as long as it is the right kind.

Common questions about matching

Q: How long does matching take? A: It depends on the pool, usually 24 to 72 hours. We would rather you wait for a good match than rush you into a bad one.

Q: Can I rematch if the first one does not work? A: Yes. Each match stands on its own. Try at least three conversations before you decide, because first impressions in text are famously unreliable.

Q: Is this like Myers-Briggs? A: No. That is a personality parlour game. Our dimensions come from behavioural research and natural-language patterns, not self-reported labels people tend to game.

My take: the next decade of social tech belongs to platforms that read psychology, not preferences. Your Spotify Wrapped will not tell you who you will bond with, but your conflict style and your sense of time might. We spent twenty years matching people on what they consume. It is probably time to match them on how they connect. Try it, or do not. But if your last match was built on both liking the same sitcom, you already know how that one ended.