How we calculate your duo.

In one sentence: your answers are placed on scales, Duolya measures the gap between you question by question using fixed rules, then writes the result from hand-written texts. No artificial intelligence calculates or invents anything.

1. Eighteen questions, six dimensions

Each person answers the same 18 questions alone, three per dimension:

• Communication: talking things out hot or cold, frankness, what makes you feel loved.
• Closeness: presence and space, texting during the day, affection in public.
• Tension: first reflex in a disagreement, pace of making up, what hurts.
• Rhythm: planning or improvising, handling surprises, organising daily life.
• Values: freedom and “us”, trust, social life.
• Future: money, where to live, children.

Most questions offer four answers ordered along a scale (from “right away” to “with time”, for instance). Two questions offer unordered choices (what makes you feel loved, what hurts you in an argument).

2. Not all differences are equal

This is the heart of the method. A gap doesn’t mean the same thing on every topic, so Duolya uses three rules:

Similarity

For most questions, being alike matters. Same answer: 100%. Answers at opposite ends of the scale: 0%. In between, the score drops steadily with the gap.

Complementarity

For rhythm (planning or improvising, reacting to surprises, organising daily life), being different can be an asset: one secures, the other opens. A gap weighs much less. Two opposite answers still keep 55%.

Sensitive divergence

For trust, money and children, a gap counts more, because these topics can’t be settled by sharing roles. Two steps apart bring the question down to about 23%, and opposite answers to 0%.

For the two unordered questions, the same answer is worth 100% and a different one 55%: not sharing the same “love language” isn’t an incompatibility, but it’s worth knowing.

3. From detail to overall score

A dimension’s score is the average of its three questions. The overall score is the average of the six dimensions, with slightly more weight for Values (×1.25) and Future (×1.2), slightly less for Rhythm (×0.9). The others count for 1.

There is no floor and no bonus. A very different duo can score 40%. We don’t inflate the score to please you: it would become worthless.

For each dimension we show a level rather than a percentage: In tune (78% and above), Nuanced (56 to 77%), Contrasting (below; Complementary for rhythm). With three questions per dimension, a single answer can shift the number by 30 points: a precise percentage would give a false sense of accuracy.

4. The duel doesn’t change the score

Each of you bets on five of the other’s answers. The number of right bets is shown separately: it measures how well you read each other, not how alike you are. You can be very different and know each other perfectly, or the reverse.

5. Why no AI calculates the score

A score must be reproducible (the same answers always give the same result), explainable (this page) and neutral (nobody tweaks it). The result texts are written by hand, question by question; the rules pick which one to show and insert your first names. Nothing is invented about you.

6. What the score can tell you

• Which topics your answers match on, and which they diverge on.
• What patterns those gaps can create day to day. For example: the more one person seeks to talk, the more the other may seek space. Couples therapists have long described this as the “demand-withdraw” pattern.
• Which conversations are worth having.

7. What it can’t tell you

• Whether you’re “made for each other”. Nobody can tell that from 18 questions.
• Whether your relationship will last. Many strong couples have modest scores.
• Who is right. No answer is better than another.
• Anything medical or psychological. Duolya is not a validated psychometric test or a diagnosis.

Your answers are self-reported and dated: they show how each of you sees yourself today. Take the duo again in a year; the result may have shifted, and that’s interesting too.

Start together →