Estimated US monthly searches for each keyword, from Ahrefs. Range in our dataset: 10 (“dog food for coat health”) to 21,000 (“fresh dog food”).

The top five keywords account for roughly 60% of total category volume. The bottom 30 keywords combined account for less than 12%. This isn’t a fat tail — it’s a fat head.

Coral = food type, navy = condition, gold = breed-specific.

Pet food’s volume distribution is unusual in one way: condition queries hit volumes you don’t see in most ecom verticals. “Dog food for kidney disease” gets 1,900 monthly searches. The mattress equivalent — “mattress for back pain” — would be a niche term. In pet food it’s a six-figure-revenue keyword. We’ll come back to this.

Ahrefs’ Keyword Difficulty score (0–100), estimating how hard it is to rank in the top 10 based on backlink profiles of current results.

KD in our dataset ranges from 0 (most condition and breed queries) to 35 (“fresh dog food”). At face value, this is a category where almost everything looks “easy.”

It isn’t. Look at “dog food for kidney disease” (KD: 0, volume: 1,900). The #1 e-commerce result is a DR 22 brand (Wynwood) ranking with a single product page. KD says zero competition. Reality says you need to out-execute a brand that has built a sophisticated schema implementation, vet-formulated positioning, and 100+ keywords ranking on a single product URL.

Out of the top 10 organic results on each SERP, how many are e-commerce pages (retailer category pages, brand product pages) vs. informational content (Reddit, vet blogs, DogFoodAdvisor, AKC). We exclude the ad slots, AIO, and product carousels — this is purely organic blue links.

Three keywords in our dataset have no e-commerce results in the top 10 at all: “dog food for upset stomach,” “dog food for constipation,” and “dog food for corgis.” The top 10 is entirely Reddit, vet blogs, DogFoodAdvisor, and Purina’s breed-info hub.

The Domain Rating distribution of the e-commerce sites ranking on each SERP. Highest DR is the strongest competitor. Lowest is the weakest one that still ranks. Median is the middle of the pack — usually the most informative number.

Median DR varies wildly by archetype:

The number of distinct organic shopping carousels that appear in the SERP — typically 0, 1, 2, or 3 per page. This is Google’s product carousel feature, populated from Merchant Center feeds, not paid ads.

On a typical food-type SERP, two product carousels appear: one near the top (above many organic results), and one mid-page. Each carousel pushes 7–10 organic results below the fold. The math: your category page often loads at position 5 even if it ranks #2 organically.

Carousels disproportionately favor brands with breed-, life-stage-, and condition-tagged products. Google’s product taxonomy reads structured data and product titles; products with explicit “puppy,” “senior,” “small breed,” or condition modifiers in their feed get pulled into more carousels than generic SKUs. Audit your product titles for these modifiers.

Whether an AI Overview appears at the top of the SERP. Binary in our dataset: 1 = present, 0 = not present. But this is the column where the data tells less than half the story — what matters more is which sources the AIO actually cites, and that’s a different game from blue-link ranking.

AIO presence breaks cleanly along archetype lines:

The pattern is consistent: the more “advice-shaped” a query is, the more likely Google is to summarize it for the user before they ever see your link. Buying queries (“wet dog food”) still get the traditional ten blue links treatment. Health-adjacent and recommendation-shaped queries don’t.

But the bigger pattern across our scrape and across our agency engagements is this: roughly 95% of AIO citations come from content that’s been updated in the last 10 months. AI Overviews aren’t pulling from your seven-year-old cornerstone article. They’re pulling from whatever was modified, expanded, or republished most recently.

Does the LLM understand what your brand is? Brands with disambiguated entity records — Wikidata entries, Wikipedia pages, Google Knowledge Graph entries, founder/exec named entities surfaced consistently across the web — get pulled into answers more often. Pet food has a specific opportunity here because most DTC brands aren’t entity-resolved at all. Royal Canin is a strong entity. “Wynwood Renal” is an emerging one. Half of DTC pet brands we audit have no Knowledge Graph presence.

When AKC.org mentions your brand alongside “best dog food for huskies,” that’s a co-citation event. When VCA Animal Hospitals references you in a piece on canine renal nutrition, that’s another. LLMs weight co-citation signals heavily because they encode third-party trust at the entity level — exactly what generic ranking signals don’t capture. Five citations from authoritative pet/vet domains can move the needle more than fifty backlinks from generic sites.

Volume matters, but so does venue. The hierarchy in pet food roughly: peer-reviewed veterinary journals > FDA / AAFCO regulatory databases > AKC, AVMA, VCA, Cornell Vet, Tufts Veterinary > mainstream press (NYT pets vertical, WSJ) > established pet-trade publications > influencer / dog-mom blogs. Mention count is a metric. Mention venue is a multiplier.

The 95% / 10-month finding holds across multiple verticals we’ve measured. Retrieval-augmented LLMs weight recency aggressively because they’re built to surface current information. The “set and forget” SEO content strategy is dead for AI visibility. Money pages need a rolling refresh cadence — re-publish dates updated, paragraphs rewritten with current data, examples swapped for current ones. We run this on a 10-month cycle for clients in fast-moving spaces, longer for slower ones.

FAQPage, Article with author, MedicalCondition, Product schema with audience properties: these all feed entity extraction. They’re necessary but not sufficient. The brands winning AIO citations in pet food are the ones doing all five vectors at once, not the ones over-investing in schema while ignoring co-citation and freshness.

Whether a “People Also Ask” box appears on the SERP. Binary: 1 = present, 0 = not present.

PAA is on essentially every pet food SERP we audited — 53 out of 54 keywords have it present. The exception was a single keyword with too little volume to generate the feature at all.

That makes the more interesting analysis the question topics themselves. Across our 54 SERPs, PAA questions cluster into a predictable taxonomy:

• Comparison — “What’s the best food for X?”
• Avoidance — “What foods should X dogs not eat?”
• Quantity — “How much should I feed?”
• Safety — “Is X safe for puppies?”
• Behavior — non-food breed and condition questions Google groups in
• Diagnosis — “How do I know if my dog is allergic?”

PAA in pet food has a strong “safety + dosing + transition” cluster that doesn’t exist in most categories. Pet food is one of the only ecom categories where shoppers actively ask “how do I switch foods” — and where the answer (“transition over 7–10 days”) is a content slot most brands leave empty. Filling it costs nothing and earns AIO citations on dozens of related queries.

The Domain Rating of the #1 e-commerce result on each SERP — i.e., the strongest e-comm site that’s actually winning, not the strongest one in the field.

Three distinct tiers emerge:

The challenger tier is the one most worth studying. Wynwood at DR 22 ranks #1 for “dog food for kidney disease” (1,900 vol). Muenster at DR 37 ranks #1 for “best dog food for huskies,” “german shepherds,” and “pit bulls.” These aren’t accidents — they’re documented patterns we’ll dig into in the teardowns.

The number of unique referring domains pointing to the #1 e-commerce page (not the whole site), and the total backlink count to that page. Two different metrics, often confused.

Referring domains range from 0 to 147. Backlinks range from 0 to 1,206. Crucially, the relationship between the two is much weaker than people assume.

Open Farm’s 86 referring domains beat Purina’s 10 — even though Purina has more total backlinks. One link each from 86 different sites is a stronger authority signal than 1,200 links from 10 sites.

The total number of keywords each #1-ranking page ranks for in the top 100 organic positions. Not just the target keyword — every keyword.

Range across our scraped pages: 5 to 1,152. The distribution maps cleanly to page format:

Estimated monthly organic search traffic to the #1-ranking page. This is the actual visit count, not the target keyword’s volume.

The headline target keyword almost never delivers most of the page’s traffic:

Volume of target keyword vs. actual page traffic

Lighter bar = volume of headline keyword. Solid bar = actual monthly visits the page receives (the “halo”).

Honest Kitchen’s human grade page brings in 4× the volume of its target keyword. That extra traffic is from the long-tail halo of related queries the page also ranks for.

For each #1 e-commerce page we scraped, we classified it as a category page (collection / filter / hub) or a product page (single SKU detail). A few are hybrid — single product page with category-level content layered in.

Wynwood is the most interesting case in our dataset. It’s a single-SKU product page (renal-chicken-pumpkin) ranking #1 for “dog food for kidney disease” — a category-level query. They’ve turned a product page into a category-level winner by giving it category-level content depth and one of the most sophisticated schema implementations we’ve seen at any DR.

The full heading hierarchy of each scraped page. Total count, distribution across levels, and whether the target keyword appears in the H1 and key H2s.

Heading patterns vary radically by format. The Chewy “high-protein dry dog food” page has 43 H2s — most of them product names — and a structured FAQ block with H2s for the FAQ wrapper and H3s for individual questions. The Royal Canin “Boxer Adult Dry Dog Food” page has 6 H3s organized around product benefits and nutritional information. Same retailer category, same archetype, completely different heading philosophies.

One pattern is universal across winners: the target keyword appears verbatim in the H1. “Sensitive Skin & Stomach Dog Food” (Purina), “High-Protein Dry Dog Food” (Chewy), “Limited Ingredient Dry Dog Food” (Chewy), “Dry Dog Food” (Open Farm). Even Royal Canin’s “Boxer Adult Dry Dog Food” has the breed in the H1.

Total visible body copy on the page, broken into three buckets: text on product pages, text before products on category pages, and text after products on category pages.

Word counts on #1-ranking pages range from 100 to 11,475. Both ends rank #1 in their respective SERPs.

The JSON-LD schema implementations on each page. We catalogued every schema type, every nested entity, and the depth of the implementation.

Schema sophistication doesn’t track Domain Rating. It tracks intent. Wynwood (DR 22) has one of the most sophisticated schema implementations in the entire dataset.

Schema stack depth across the field

Distinct JSON-LD @type entities present on the #1-ranking page.

The Wynwood stack:

Compare to Royal Canin: a basic Product + BreadcrumbList + Organization stack. No FAQPage. No AggregateRating on many breed pages. No MerchantReturnPolicy. Purina — DR 80, sitting on enormous link equity — does even less. They rank because of brand, not because of structure.

The title tag and meta description on each scraped page, with character counts and keyword analysis.

Title length runs from 14 to 70 characters. The variation is wild:

Royal Canin uses the same 14-character title — “Royal Canin US” — across every breed-specific product page we scraped. Their meta descriptions on those pages are mostly missing entirely. And they still rank #1 for nearly every breed-specific query.

Across our scraped winners, the title formula that consistently appears is: [Target Keyword] | [Brand] — sometimes with a benefit modifier in between. Chewy’s “WEIGHT MANAGEMENT DOG FOOD – WEIGHT LOSS DOG FOOD | CHEWY” is on-the-nose but works. Open Farm’s “All Dry Dog Food & Ethically Sourced Dog Kibble | Open Farm” adds the brand differentiator in the title itself.

Internal links from the body content of each scraped page (excluding navigation and footer). Anchor text and target.

Across roughly 30 scraped pages, “There is no internal links” appears as the audit result on the majority.

Body internal links per #1-ranking page (excluding nav & footer)

The exceptions are loud. Muenster’s pit bulls page links to dog food pages for huskies, German shepherds, labs, goldendoodles, and yorkies — a literal breed-page mesh. Tractor Supply links to “Senior Dog Food,” “Affordable Dog Food,” and other related Tractor Supply category hubs. Open Farm’s “natural dog food” page links to “natural cat food” and a few related categories. Everyone else: zero.

Whether the #1-ranking page is reachable directly from the site’s main navigation menu (header), and the anchor text used.

Most #1-ranking pages we scraped are not in the main nav. The pages that are, tend to be brand-level hubs (Purina’s “Senior Dog Food,” “Sensitive Stomach”). Filter URLs (Chewy’s /f/low-fat-dry-dog-food_c294_f3v643879) almost never make the nav, which is fine — they’re long-tail capture pages, not architectural anchors.

The interesting cases are mid-DR brands. Hill’s links “Dog Food” from the nav. The Honest Kitchen links “shop dog food” as a direct nav anchor. Open Farm doesn’t put their “dry dog food” page directly in the main nav — it lives one click deep behind “Dog Food.”

Whether visible breadcrumb navigation appears on the page (not just in BreadcrumbList schema).

Almost every page we scraped has BreadcrumbList schema. Most of them don’t have visible breadcrumbs on the page itself.

The pattern is so consistent it’s practically a category-wide default: implement BreadcrumbList in JSON-LD for the rich-result benefit, then leave it out of the visual UI. The exceptions (Wynwood, Open Farm) tend to be brands where SEO and design are coordinated rather than handed off to separate teams.