We checked our scores against an independent expert. The cat scores didn’t agree.
330 products, two species, one uncomfortable result.
8 September 2026
FurMetric gives dog and cat food a score out of 100. The obvious question is whether that number means anything, and the honest answer is that we couldn’t know from the inside.
Our scoring rules were calibrated against three products — a premium food, a mainstream one, and a grocery-tier one. That tells you the model is sane on three products. It tells you nothing about the other fourteen hundred.
So we checked it against somebody else. Dog Food Advisor is the most-read independent voice in the same tradition we score in: named meats count for a lot, unnamed by-products and cheap fillers count against. If our rules generalise, we should broadly agree with them. If we don’t, we should find out before a customer does.
The dog result
0.785 rank correlation across 16 product lines, covering about 200 scored dog foods.
That is close agreement. More usefully, the extremes land almost exactly where they should: the food they rate worst, we score between 19 and 37. The food they rate best, we score 95. Nothing was wildly out of place anywhere in between.
Where we do something they can’t
Six different premium lines all sit at the top of their five-star scale. We spread the same six products from 86 to 95.
That isn’t a disagreement, it’s a difference in resolution. Five stars is a ceiling — once several foods hit it, the scale stops telling you which to buy. A 0–100 score keeps working above that point. If you’re standing in an aisle choosing between two bags that both look excellent, that gap is the entire decision.
The cat result, which is the interesting one
0.594 rank correlation across 12 lines, covering about 130 scored cat foods.
Noticeably weaker. And with only twelve lines, the gap between the two numbers isn’t statistically solid — we could be reading noise.
Except the disagreements aren’t scattered. They all run the same direction, which is what makes them worth taking seriously:
- We are harsher on the veterinary-brand cat foods. The ones with ordinary ingredient lists — by-product meal, corn — but strong, carefully formulated nutrient profiles. The independent reviewer credits the profile. We mark down the list.
- We are more generous on boutique cat food. Grain-free kibble with a beautiful ingredient list that is still, when you do the arithmetic, quite high in carbohydrate. They mark that down. We largely don’t.
Our best guess at why — and what we did about it
Our scoring weights ingredient quality at 40% and nutritional adequacy at 20%. Those are the same weights we use for dogs.
Cats are obligate carnivores. Dogs aren’t. Applying one balance to both species is arguably wrong, and every single divergence lands exactly where that mistake would put it: foods with plain ingredients but excellent nutrition get marked down by us, and foods with lovely ingredients but mediocre nutrition get marked up.
That is a tidy explanation, which is exactly why we haven’t acted on it. Twelve product lines is not enough evidence to change how an app scores food. We have written the hypothesis down, and we will re-run the comparison when the cat catalogue is large enough to give a real answer. If the gap survives, we will change the weights and say so.
Two things we did not do
We did not copy anything. No reviews, ratings or ingredient lists were imported into FurMetric from anyone. This was reading a public website once, to check our own homework. Their ingredient lists are transcribed off physical labels exactly like ours, so copying them would not even have helped.
We did not quietly fix the disagreements. It would have been easy to nudge the cat weights until the correlation looked better, and it would have meant nothing — tuning a model to match an answer key is not the same as being right.
Why publish the half that looks bad
Because an app that scores things should be able to say where its scores are least certain. Our dog scores track an independent expert closely. Our cat scores are more of an open question, and you deserve to know that before you trust one.
We re-run this check after any change to the scoring rules. It costs nothing and it is a far better test than three products ever was.
FurMetric scores dog and cat food out of 100. Scan the barcode on any bag and see the score with the reasoning behind it — which ingredients earned credit, which lost points, and why.
Unlimited scans, always free. No account. No ads, no sponsored placement, and no affiliate links — a brand cannot buy a better score.