How Love Compass works
Love Compass estimates how common your preferences are among adults worldwide. This page explains, in plain language, what is calculated, what it is based on, and where it stops being reliable.
We have tried to state only what we can support. Where something is an estimate, an assumption, or not independently verified, we say so.
What the assessment measures
It estimates the size of a group: roughly how many adults, out of everyone in the world, would meet every criterion you choose.
That is all it measures. A bigger group means your preferences are common. A smaller group means they are less common. It is a statement about preferences, not about you or about any person.
Two different calculations
The free quick check and the paid full report work differently. We keep them separate here so you know which one you are looking at.
Free quick check (Reality Check)
A fixed calculation anyone can follow. It uses three questions: gender, minimum income, and education.
The result is the same every time for the same answers. Nothing is random.
Full report
The full report covers more criteria (for example age, location, religion, height, lifestyle). Its numbers are produced by an AI model that is given a written brief with rounded worldwide reference figures and is asked to apply each of your criteria as a narrowing step.
This means the full report is not a fixed formula. It is an AI-generated estimate, so the same answers can produce somewhat different numbers on a different run. We have not published an accuracy test of it.
Which inputs affect the estimate
Free quick check
- Who you are looking for (a man, a woman, or no preference)
- A minimum yearly income
- A minimum education level
Age, location, religion, height, lifestyle and personality are not part of the quick check's number.
Full report
- Everything above, plus age range, location, race, religion, profession, height, body type, marital status, children, and personality traits you select
- Your own zodiac sign, used only for a separate written zodiac insight, not for the pool size
Choosing "no preference" for a criterion means it does not narrow the group.
How filters and overlapping criteria are handled
In the quick check, each answer keeps a share of the world's adults, and the shares are multiplied together. For example, an answer that keeps 51% of adults, then one that keeps 20%, then one that keeps 25%, leaves 51% × 20% × 25% = 2.55% of adults (shown step by step in the example below).
Multiplying like this treats each criterion as unrelated to the others. In real life they overlap: people with higher education tend also to have higher income, for instance. Because of that overlap, the true group can be larger or smaller than the multiplied figure. We do not adjust for this.
In the full report, the AI model is asked to narrow the group one criterion at a time. How it handles overlap is not a documented, fixed rule, and we cannot tell you it corrects for overlap accurately.
A worked example from the quick check
These steps are computed live by the same calculation the quick check uses.
Example answers: a woman, $30k+ income, bachelor's degree or higher.
- Start: all adults worldwide (about 5.2 billion): 5.2B
Step 1
Looking for a woman
keeps 51.0% · pool after this step: 2.7B
Step 2
Income of $30k or more
keeps 20.0% · pool after this step: 530M
Step 3
Bachelor's degree or higher
keeps 25.0% · pool after this step: 133M
Result
About 133M adults, or 2.6% of all adults.
We show this as a band of roughly 119M to 146M.
Data sources, dates and geographic scope
Geographic scope
Worldwide only. The quick check does not look at any country, region or city. The full report can name places where matching people may be more common, but those place suggestions come from the AI model, not from a local dataset we hold.
The reference figures used
- About 5.2 billion adults (18+) worldwide
- About 49% men and 51% women
- Income: about 55% earn under $10k a year, 25% earn $10k to $30k, 12% earn $30k to $75k, 6% earn $75k to $150k, 2% earn over $150k
- Education: about 55% have no degree, 20% some college or vocational study, 15% a bachelor's degree, 10% an advanced degree
What we have not verified
- These figures are rounded approximations. The app does not record which survey or publication each one came from, or what year it describes. We therefore cannot give you a precise citation or date for them.
- We have not independently checked them against the sources listed below. Treat them as rough orders of magnitude, not as official statistics.
- "Income" is not defined in the figures (for example personal or household, or which currency and year). It is a simple tier, not a measured income.
Where to check real figures yourself
These are respected public sources on the topics involved. They are listed so you can look up current official figures. They are not a claim that our numbers were taken from them or match them.
Assumptions and estimates where direct data is missing
- Rounded shares: the reference figures are rounded to whole percentages, so small differences are lost.
- Independence: criteria are multiplied as if unrelated (see above).
- Worldwide averages: a global figure is applied to everyone, even though income and education vary hugely between countries.
- Plain tiers: "$30k+" and similar are simple cut-offs and do not account for cost of living or currency differences.
- Rounding band: the quick check shows a band of about plus or minus 10%. This is a rounding allowance we chose. It is not a statistical confidence interval and it does not measure real uncertainty.
- AI estimates: in the full report, the numbers, place suggestions, per-place scores and written tips are generated by an AI model from the brief it is given. They are estimates, not measurements.
- Fixed what-if multipliers: the "what-if" scenarios in the full report use preset multipliers built into the app (for example, relaxing an income tier multiplies the pool by a set factor). Each carries a source label, but we have not verified that the multiplier was derived from that source. Treat them as rough illustrations.
What the realism label means
The label (Realistic, Selective, Highly selective, Statistically rare) is based only on the share of adults who fit your answers. It is not a judgement about you or your preferences.
These cut-offs are our own choice of bands. They are not taken from a published standard.
What the results cannot tell you
A pool size describes how common a set of preferences is. It cannot predict:
Chemistry
Whether you would click with someone is not something demographics can measure.
Mutual attraction
People who fit your criteria also have their own preferences. Nothing here estimates whether they would be interested in you.
Relationship success
A match on paper says nothing about whether a relationship would be happy or lasting.
Finding a life partner
The number is not a probability that you will meet or marry anyone. Where you live, who you meet, timing and effort all matter, and none of them are in the calculation.
Your own value or desirability
The result describes your preferences, not you.
Some screens in the full report use the word "probability". In every case this means the estimated share of adults who fit your criteria. It does not mean your chance of finding love.
How to interpret your result responsibly
- Read it as "how common are these preferences?", not "how likely am I to find someone?"
- Treat the number as an order of magnitude. A difference between, say, 9% and 11% is not meaningful.
- A small group is not a verdict. It can simply show where relaxing a criterion would widen your options the most.
- Do not use it to judge other people, or yourself. Criteria describe preferences, not worth.
- Use it to start reflection, not to make a major life decision.
- Remember it is worldwide. Your own area may look very different.
Validation and accuracy
We have not carried out or published a validation study of these estimates, and we do not claim any accuracy rate, success rate or comparison with real dating outcomes.
If we later test the method against independent data, we will describe the results here.