How it works

Every trait starts as “not important”. That is not “medium importance” or “a weak requirement” — the trait is switched off entirely and moves the breed neither up nor down.

So the matcher never invents requirements for you. If you said nothing about barking, it will not decide on your behalf that a quiet dog is better: quietness becomes no advantage, and a loud voice no fault.

The far end of the scale — “required” — works as a filter, not as a heavy weight. For wanted traits it means “a high level is required”, for unwanted ones “a low level is required”, and the form always names the direction: “Low barking tendency — required”. A breed that breaks such a requirement moves into a separate “Excluded by hard requirements” block with the reason named, rather than sinking to the bottom of the list.

The difference matters: a heavy penalty can be outweighed by strengths elsewhere, and a breed with a deal-breaking trait would still surface near the top. Exclusion does not allow that — but it does not hide anything either: the reason is visible and the list unfolds.

The percentage is the share of your requirements that are met, weighted by how important each one is to you. Within a trait the match is not yes-or-no but a degree: a rating of 2 against “undesirable” counts partially rather than scoring zero.

Only the traits you marked are counted. A breed earns no points for qualities you never asked about.

Weight, height and size category work as a filter: breeds outside your range do not appear at all. Inside the range every breed is equal — no percentage is awarded for “fitting the size”.

Otherwise a dog in the exact middle of your range would appear to “fit better” than one near its edge, although both suit you equally.

Behavioural ratings are not filled in for every breed in the catalogue. An unrated breed gets no percentage and appears in its own block: it passes the physical limits, but there is nothing to compare it on.

A partially rated breed has one more condition: it enters the ranking only if at least half of the traits you marked are rated for it. Without that rule a breed with two filled cells would score a hundred per cent and come first — simply because almost nothing is known about it.

If no breed reaches the threshold, the matcher does not pass weak results off as matches. It says plainly that no breed fits, and separately shows the closest alternatives together with the requirements they fail.

More often than not this means your requirements conflict with each other, not that no dog exists for you.

Every rating on a breed profile carries a source and a confidence level. Some values are expert judgement and need verification; where a rating rests on research or on data from a kennel organisation, that is visible.

Next to the number there is a note explaining what produced that score. This is not decoration: the Beagle and the Husky both score high on roaming, but for completely different reasons — one follows a scent trail, the other simply runs. What they need from an owner differs accordingly.

Ratings describe a breed statistically. According to Morrill et al. (Science, 2022), breed explains about 9 % of the variation in behaviour between individual dogs. The rest is bloodline, upbringing, experience, health and plain individuality.

That does not make matching pointless: 9 % is a noticeable amount when it comes to roaming or barking. But read the result as a probability, not a promise. No matcher replaces meeting the actual dog and talking to the breeder.