Self-reported preferences are a terrible way to find a partner. I spent 2026 testing what works instead.
The Question
Every dating app asks you what you want. Tall, funny, ambitious, likes dogs. You fill in the fields and the algorithm finds someone who matches your answers.
The problem: people are bad at knowing what they want. They describe their ideal partner based on who they think they should want, not who they actually respond to.
The Experiment
The One was a behavioural matchmaking experiment. Instead of asking participants what they were looking for, it observed how they responded - to scenarios, to tension, to ambiguity, to other people's decision-making patterns.
The hypothesis: revealed behaviour predicts compatibility better than stated preferences.
What Happened
The experiment ran across India through June 2026. It drew a pool that skewed 75% male, and the algorithm computed 350+ matches from behavioural inputs alone. Along the way it forced careful thinking about product design, survey architecture, behavioural data, and what a matchmaking system actually needs to do to work.
The One concluded in June 2026, with a closing note sent to every participant.
Where It Lives Now
The research continues under The One Labs, the behavioural research entity that grew out of the experiment. The dataset, the patterns, and the questions it raised are the raw material for what comes next.