How FurScore Works: The Algorithm Behind Every Rating

When we set out to build Furra, one of the first questions we had to answer was: how do you score a dog food fairly? The market is huge, the formats wildly different, and every brand has its own way of presenting information. We needed something consistent, repeatable, and free from human bias. That's FurScore.

FurScore is an algorithm that scores every product on Furra out of 100. It runs automatically, it treats every food by the same rules, and nobody on the team can adjust a score manually. Once the algorithm runs, the numbers are what they are.

The Dry-Matter Problem

Before any scoring could happen, there was a problem to solve first. Every manufacturer publishes their nutritional data in as-fed figures, meaning the numbers reflect the food as it comes out of the bag or tin, moisture included. A wet food might be 78% water. A kibble might be 8%. If you compare protein percentages between those two foods as-fed, you are not comparing the same thing at all. You are mostly comparing water content.

So the first thing the algorithm does is convert every product to a dry-matter basis, stripping out the moisture to leave only the actual nutrients. Only once everything is on the same footing can you make a fair comparison. It sounds simple, but getting this right across 2,500 products in every format was the foundation everything else was built on.

The Five Pillars of FurScore

The score is made up of five pillars. Nutrition is the biggest, worth up to 60 points. It looks at meat content, protein levels, carbohydrate levels, and ash content. Higher meat and protein push the score up. Higher carbs and ash push it down. We use dry-matter figures across the board so that wet food, kibble, raw, and fresh formats can all be compared on a level playing field.

Ingredient Quality

Ingredient quality accounts for 15 points. Every product starts with the full 15, and points are deducted for each vague or controversial ingredient in the list. Things like 'meat and animal derivatives', 'cereals', or 'oils and fats' each cost points. Named, specific ingredients are rewarded simply by not being penalised.

Value, Traits, and Processing

Value for money is worth 10 points, based on the cost per 100g. Traits and labelling add up to another 10 points, covering things like clear labelling, natural ingredients, and whether the food is nutritionally complete. The final 5 points come from format and processing. Raw and fresh foods score highest. Extruded kibble scores lowest, not because kibble is bad, but because minimal processing is generally a positive signal.

I wanted the algorithm to reflect what we actually know about dog nutrition, not what brands want you to think. If a food hides behind vague ingredient terms, the score should reflect that. / Niko Moustoukas

One thing that makes FurScore different from a fixed star rating is that it is relative and percentile-based. Every product is ranked against the entire pool of enabled products on the site. That means if a genuinely excellent new food is added, it can push others down slightly. If a poor product is removed, others may rise. The scores reflect the current state of the market, not a fixed scale set years ago.

FurScore runs in two situations. It runs whenever a batch of new products is added to the platform, and it runs once a month to keep everything fresh and accurate. Between those runs, scores remain stable.

Every product on Furra has a score. Every score is calculated the same way. That's the whole point.

FurScore is also never really finished. As our understanding of dog nutrition develops, as we add new data points, and as the market changes, the algorithm will change with it. We'll always be transparent about significant updates when they happen.