How We Grade Dog Food

Last Verified: May 25, 2026 · Author: Matt Gehring, Sniff founder · Full technical specification →

Why this exists#

There are over 7,000 commercial dog food products on the U.S. market. The average pet food aisle has 200+ options. And the person standing in that aisle, the one who just wants to feed their dog well, has almost no way to tell the difference between a genuinely good formula and a bag with great marketing.

That's not their fault. The information exists. It's buried in veterinary journals behind paywalls, locked inside regulatory documents written for industry insiders, and scattered across PubMed abstracts that nobody outside a university library will ever read. The people who know the most about canine nutrition, board-certified veterinary nutritionists and researchers at places like Cornell and Tufts and UC Davis, publish extraordinary work. It never reaches the person who needs it most.

Meanwhile, what *does* reach them is ingredient-panel ideology. "Chicken is first, so it must be good." "Grain-free is healthier." "Byproducts are bad." These aren't conclusions drawn from evidence. They're marketing narratives that have been repeated so often they feel like facts.

Sniff exists to close that gap. We've read the labels on over 3,500 dog food products, scored every one against the same rubric, and published every step of the math. Every number on this site traces to a published source. Every source is linked. If we're wrong, you can see exactly where and tell us.

This page explains how every Sniff Score is computed. No black box. No secret formula. If you disagree with a score, read this page, identify what you think we got wrong, and let us know. We'd rather have that argument in the open than pretend our number is the final word.

The two things we measure#

Every product on Sniff gets two ratings, not one. They always appear together because neither tells the full story alone.

The Sniff Score: 0–100

This is the nutrition and quality grade. It answers: how good is this food's actual formulation?

We look at the protein sources and how digestible they are. We look at the fat quality and whether the omega-3s come from real marine sources or just flaxseed (dogs convert less than 1% of plant-based ALA to the DHA they actually need). We look at the carbohydrate quality and whether the formula relies on ingredients linked to ongoing safety investigations. We check AAFCO compliance, controversial ingredients, and whether the guaranteed analysis actually matches what the label promises.

All of that gets compressed into a single number from 0 to 100.

These bands were calibrated against the full product catalog. They aren't arbitrary. They reflect the actual distribution of quality in the commercial dog food market.

The Evidence Rating: how much we trust our own score

This is the part most review sites skip. We don't just score the food. We tell you how confident we are in that score.

Five levels: Robust · Good · Fair · Limited · None.

A food scoring 88/A with "Limited" evidence means the formulation looks excellent, but we couldn't fully verify the data. Maybe the manufacturer doesn't publish feeding trial results. Maybe the guaranteed analysis came from a retailer listing rather than the brand itself.

A food scoring 55/C with "Robust" evidence means this is a thoroughly documented, nutritionally average product. We're very confident in the score, and the score says it's mediocre.

The Evidence Rating exists because large manufacturers have an inherent advantage in data availability. Companies like Purina, Hill's, and Royal Canin can afford board-certified nutritionists on staff and formal feeding trials. That doesn't automatically make their food nutritionally superior. It means we have more data to verify their claims. A small brand with excellent ingredients might score A on nutrition but Limited on evidence. That's a real signal, not a flaw in the system.

What goes into the score#

The Sniff Score is built from nine components organized into three tiers. Here's what each one measures and why it matters, in plain language. The exact weights and formulas live in the full technical specification for anyone who wants to audit the math.

Tier 1: Is the nutrition actually good?

This is the bulk of the score, and it should be.

Protein quality is the biggest single factor. Protein is the foundation of canine nutrition. We don't just check whether "chicken" appears on the label. We evaluate the digestibility and biological availability of every protein source in the first ten ingredients, using the most rigorous canine protein assay available (Templeman & Shoveller 2022). Named whole meats and same-species meal combinations score highest. Vague labels like "meat meal" or "poultry byproduct meal" score lowest, not because byproducts are inherently bad, but because you deserve to know what species is in the bag.

Fat quality matters more than most people realize. We reward named fat sources and penalize vague ones ("animal fat" could be anything). Most importantly, we check whether the omega-3 fatty acids come from marine sources like fish oil, salmon oil, or algae, because that's where EPA and DHA actually come from. Flaxseed provides ALA, which sounds good on a label, but dogs convert almost none of it to the DHA their body needs. If you want to understand why this matters, our guide on reading dog food labels breaks it down.

Carbohydrate quality is where we make a distinction almost nobody else in the industry makes: grain-free and high-legume are not the same thing.

A grain-free formula built on sweet potato or tapioca with no pulses? No penalty. That's a legitimate formulation choice.

A grain-free formula loaded with peas, lentils, and chickpeas in the top five ingredients? That's a WATCH-level concern. The FDA investigated a potential link between these formulations and a heart condition called dilated cardiomyopathy (DCM) from 2018 to 2022. Recent biomarker research has identified a plausible biological mechanism, though the work still needs replication. The science isn't settled. The signal is strong enough that we'd rather be cautious. The full picture is in our guide on grain-free food and DCM.

Tier 2: Is it safe and transparent?

AAFCO compliance isn't just a checkbox. *How* a food meets AAFCO standards matters enormously. A feeding trial, where real dogs ate the food for six months and were monitored, is real evidence. "Formulated to meet AAFCO nutrient profiles" is paper compliance: a computer confirmed the recipe *should* work, but nobody tested it. No AAFCO statement at all? The score is hard-capped at 59, because without it there's no regulatory baseline for nutritional adequacy.

Ingredient specificity rewards transparency. What fraction of the protein sources are species-named? "Chicken meal" tells you something. "Poultry meal" tells you less. "Meat meal" tells you almost nothing. If any unnamed protein source appears in the top three ingredients, the score takes a significant hit.

Controversial ingredients are where the scoring gets its teeth. We maintain a watchlist of 25 ingredients, and each one carries a published verdict: FLAG, WATCH, or CLEAR. Every verdict has a basis type (confirmed canine harm, regulatory action, or precautionary) so you can see exactly why we flagged it and decide for yourself whether you agree. The full controversial ingredients guide explains each one.

Recall surveillance runs continuously. We pull from the FDA Center for Veterinary Medicine recall feed every 24 hours. If a product has an active unresolved recall in the last 12 months, the score is capped at F tier regardless of how the formula looks. Most consumer review sites surface recalls as a separate alert, if they surface them at all. We bake them into the score itself, because a recall *is* a quality signal.

Tier 3: Is it nutritionally complete?

Guaranteed analysis adequacy checks whether the food actually meets AAFCO minimums on a dry-matter basis for its labeled life stage. We compute this ourselves from the declared values. We don't trust marketing claims.

Caloric density is form-dependent. A 250 kcal/cup kibble is suspiciously low. A 700 kcal/cup kibble is dangerously high. Wet food, freeze-dried, and raw each have their own appropriate ranges.

Micronutrient indicators reward formulation investment. Chelated minerals (proteinated forms) are more bioavailable than oxides. Natural vitamin E is preferred over synthetic. Evidence-based probiotics with declared CFU counts at expiration signal a manufacturer that's paying attention to the details.

The hard caps: some things are non-negotiable#

This is the most important mechanic in the entire system, and it's inspired by how Yuka scores food products in Europe.

A premium ingredient list doesn't wash away a suspected carcinogen. If certain ingredients are present, the score is hard-capped, no matter how good the rest of the formula looks.

Why this matters: a dog food could have beautiful protein sources, excellent fat quality, and perfect caloric density. But if it contains BHA, classified as a Group 2B probable carcinogen by the International Agency for Research on Cancer and listed under California Prop 65, the score cannot exceed 49. Period.

This is how the system encodes a simple principle: some risks can't be averaged away by benefits.

Three positions we take (and why)#

Scoring dog food requires making judgment calls. We'd rather make them openly than pretend we don't have a perspective.

1. Grain-free isn't the problem. Pulse-heavy is the concern.

The pet food industry, and a lot of the internet, treats "grain-free" as a single category. It's not. A sweet-potato-based formula and a pea-lentil-chickpea-based formula are fundamentally different products that happen to share a marketing label.

Our penalty applies to the pulse-heavy pattern specifically: formulas where legume ingredients dominate the carbohydrate profile. This is the pattern the FDA flagged. This is the pattern recent biomarker research is starting to characterize. We don't penalize grain-free. We penalize the specific formulation choice that the evidence questions.

We extended the detection window from the top 10 ingredients to the top 15 in version 2.0.1, because some brands were concentrating pulses at positions 11 to 15 to technically escape the rule while functionally keeping the same formula. The methodology adapts when brands try to game it.

2. Sodium selenite is fine. Copper is the real concern.

The internet pet food community is convinced sodium selenite is dangerous. The peer-reviewed literature disagrees. At AAFCO-permitted levels (max 2 mg/kg DM, typical use 0.1 to 0.3 mg/kg), there are zero documented cases of canine harm. A scoring system that flags sodium selenite is being performative, not scientific. We give a small bonus for selenium yeast as a marginal quality upgrade, but sodium selenite gets no penalty.

Meanwhile, copper-associated hepatopathy is a genuine and growing safety signal. After 1997, the pet food industry shifted from low-bioavailability copper oxide to high-bioavailability copper sulfate and copper chelates. Hepatic copper concentrations in dogs have risen significantly since. Dr. Sharon Center at Cornell has documented this extensively, and her 2021 JAVMA commentary called for immediate regulatory attention.

We classify copper as WATCH, not FLAG, because we can't verify total copper content from labels alone. Manufacturers aren't required to disclose it. When they list copper sulfate, we note the concern. When they use copper proteinate or add no supplemental copper, we assess neutrally. This is more conservative than the leading expert opinion, but it's the most we can responsibly claim with label data.

We want to do more. When AAFCO mandates copper disclosure or brands start publishing independent lab assays, we'll upgrade our position. We wrote a full guide on copper hepatopathy that explains the regulatory history, the breed predispositions, and where the science stands today.

3. Every flag comes with a receipt

Every controversial ingredient on Sniff carries a basis type so you can see exactly why we flagged it:

Confirmed harm means documented canine injury or death at commercial exposure levels. These are facts, not opinions.

Regulatory action means a major jurisdiction (FDA, EFSA, AAFCO) has banned, restricted, or issued a formal warning. The regulatory system flagged it before we did.

Precautionary means there's a plausible concern from non-canine evidence or a jurisdictional gap. Banned in the EU but permitted in the US, for example. We publish the precautionary label specifically so you know this is a judgment call, not a proven danger.

Reasonable people can disagree about precautionary flags. That's fine. We publish the basis so the disagreement can be specific rather than vague.

Every product page shows its work#

We don't ask you to take a score on faith. Every product page shows the top three things that helped the score and the top three things that hurt it. Real example:

✓ Strong protein profile. Deboned chicken as primary protein with high biological value.

>

✓ Quality fat sources. Named chicken fat with salmon oil for omega-3.

>

✓ AAFCO feeding trial substantiation for all life stages.

>

✗ Contains BHA. IARC Group 2B probable carcinogen, CA Prop 65 listed.

>

✗ No declared omega-3 marine source. Flaxseed alone cannot raise DHA levels.

>

✗ Score capped at 49 due to FLAG ingredient (BHA).

These driver statements are designed to be self-contained facts. They're also what AI assistants extract when someone asks "is this brand good for dogs?", because they're citation-ready by design.

How we're getting better#

Sniff is not a finished product. It's a versioned system that improves as the evidence base grows.

The scoring rubric is on its second major version, with multiple point releases since launch. Every change is documented in the version history. When new evidence challenges an existing position, we don't quietly adjust. We publish a changelog entry explaining what changed and why.

If you think the methodology should change, if you're a veterinary nutritionist who disagrees with a position, or a researcher whose work we've miscited, or a brand that believes we have a factual error, tell us. We respond to every submission within 48 hours and provide a written determination within 14 days.

What we don't know#

Epistemic honesty matters more than authority. We'd rather tell you what we're uncertain about than pretend we have all the answers.

We don't know what causes diet-associated DCM. The FDA received 1,382 reports and then closed routine updates, citing insufficient causal data. Recent biomarker work is starting to characterize a plausible mechanism, but it needs replication before anyone can call the question settled. Our pulse-heavy penalties are precautionary, not causal.

We don't have a perfect protein quality database for dogs. The best available data uses rooster-derived digestibility proxies. True canine-specific ingredient quality may be different than what we estimate. We use the best data that exists and we say so.

AAFCO has no senior life stage profile. Our senior-specific scoring leans on published sarcopenia research, which is extra-regulatory. We're scoring beyond what the regulatory framework requires because the science supports it.

The optimal omega-6 to omega-3 ratio isn't precisely established for dogs. Working consensus says 5:1 to 10:1. AAFCO permits up to 30:1. Wild canid diets are estimated at 1:1 to 4:1. We score toward the lower end of that range because the anti-inflammatory evidence supports it.

Copper hepatopathy is a strong signal without a regulatory response. AAFCO declined to set a copper maximum in 2023. NRC never established a safe upper limit. Our position is on the leading edge of expert opinion, not regulatory consensus. We wrote 3,000 words about why.

Starch tolerance varies dramatically by breed. The AMY2B gene, which controls starch digestion, has between 2 and 34 copies depending on breed. Saying "dogs digest starch well" is a population-level statement that may not apply to your specific dog.

If you read this list and think "they're being too cautious about X" or "not cautious enough about Y", you might be right. Tell us. The methodology is versioned, the changelog is public, and we'd rather be corrected by an expert than be confidently wrong.

Who we are. Who we aren't.#

Sniff is independent. We accept no money from any pet food manufacturer, retailer, or affiliate network. No commission, no sponsored placement, no preferential ranking. Product pages contain no commission-bearing "buy" links. The full stance lives at sniffscore.co/independence, including what would change if affiliate partnerships are added in the future: flat rate for every product, published charity split routed through the Pax Pledge, editorial always senior to commerce.

We are not veterinarians. Nothing on Sniff is medical advice. If your dog has a health condition, talk to your vet, ideally one who is board-certified in nutrition (DACVIM-Nutrition). There are fewer than 100 of them in the United States, and they are the highest authority on what your specific dog should eat.

What we are is a bridge. We read the research that veterinary nutritionists publish, we build systems that make it accessible, and we show our work so anyone, including those nutritionists, can check it. If we've gotten something wrong, the correction form is right there.

Dig deeper#

This page is the plain-language overview. For the full technical specification, every formula, every weight, every threshold, every citation, see the Technical Methodology.

To understand specific topics in depth:

• Copper in Dog Food: The Hepatopathy Guide. The regulatory gap, the breed predispositions, and where the science stands.
• Grain-Free Food and DCM. What the FDA found, what the latest biomarker work is showing, and why we take a precautionary stance.
• AAFCO Statements Explained. Why "complete and balanced" means less than you think.
• How to Read a Dog Food Label. The guide we wish existed when we started.
• Controversial Ingredients: The Full List. All 25 ingredients, every verdict, every citation.

To challenge a score, report an error, or submit new evidence: [email protected] or the corrections form.

*This page is versioned. Current version: v2.0.2. The full changelog documents every update since launch. The methodology will continue to evolve as the evidence base grows, and every change will be published here.*