How AI Assistants Are Changing Pet Product Discovery

Fresh food, functional treats, and condition-specific supplements are pulling in more of the pet industry's spending every year, while commodity products flatten out. The buyers driving that shift, mostly millennials and Gen Z owners, research before they buy, and a growing number of them start that research with an AI assistant instead of Google. This piece looks at how those assistants pick which pet brands to recommend, why the process tilts toward household names, and what a brand actually has to do to earn a spot in the answer.

Vets, word of mouth, and in-store browsing haven't gone anywhere. But a new layer has formed on top of them, and it moves fast because it does something those older channels can't: it synthesizes. Ask an AI assistant for a grain-free food for a senior dog with joint issues and a shellfish allergy, and it hands back a shortlist in one turn. No clicking through five retailer filters. That's why this channel is maturing faster than most pet marketing teams have clocked.

AI assistants don't rank brands the way Google ranks a results page. There's no crawl-and-index logic running in the background, and no ad auction either. These systems build an answer from what they learned in training, plus, on platforms that support it, live retrieval from the web and whatever sources they choose to cite. Nobody buys a placement here. A brand earns its way in through content authority and citation signals, or it doesn't show up at all.

Different platforms weigh sources differently, and that's where things get uneven. Some lean hard on editorial and earned media, pulling from sites like PetMD, The Spruce Pets, and Kinship. Others give real weight to Reddit threads where pet owners argue about kibble brands at two in the morning. Some favor long-form publications and review aggregators; others pull from video transcripts and the metadata behind visual content. Chewy's editorial arm and the veterinary trade press turn up repeatedly as feeder sources no matter which platform you check.

Run the same prompt through ChatGPT, Gemini, Claude, and Perplexity, and you get four different shortlists. Each model has a different training map and a different appetite for which sources it trusts, so a brand that dominates one platform's answers can barely register on another's for the identical query.

Here's the part that should worry marketing teams most: most AI search sessions end without the user ever clicking through to a website. The recommendation is the decision point. There's no second chance to land an impression on a product page, because the product page never gets visited.

The systematic bias toward mass-market brands — and what it costs specialist ones

AI assistants consistently put mass-market brands at the top of pet product answers, and the reason is mechanical, not editorial. These are the names with the deepest footprint in training data, the most third-party coverage built up over decades, the highest review counts on every retailer site. Models pull toward what they've seen most often, and legacy brands have simply been around longer, written about more, reviewed more.

That pull cuts against where the market is actually headed. The fastest-growing segment in pet care is premium and specialty, yet AI discovery keeps underrepresenting exactly that segment. Fresh, direct-to-consumer pet food makes the case plainly. The category is expanding fast, with some projections putting it at several times its current size within a decade. DTC leaders like The Farmer's Dog built consumer recognition the hard way, through years of subscription relationships and word of mouth. Legacy players are moving in, though. General Mills bringing Blue Buffalo into the fresh segment brings a content footprint that took decades to build. Early AI citations still favor the DTC leaders, but that contest is live, and nothing about the outcome is settled.

Pet insurance shows a quieter version of the same gap. Brands with genuinely strong market positions still see citation share lag behind where they actually stand in the category, a gap a deliberate content program could close if someone bothered to run one. Durable toys tell a sharper story: KONG owns citation share for the core dog-chew queries almost outright, but its own cat toy and bird toy lines get cited far less than the company's actual market presence would suggest. That's not a brand awareness problem. It's a content gap, and it's fixable.

Being under-cited creates a bind, and neither side of it is good. A brand can be invisible, meaning pet owners never encounter it in an AI answer at all, or it can be misrepresented, showing up with the wrong price, an outdated formula, or a health claim nobody made. In a category built on animal nutrition and medicine, that's not a small error. Both outcomes cost a brand customers, and the second one carries a safety and reputational weight most consumer categories don't have to think about.

Why the same brand looks different depending on which AI platform a pet owner uses

Platforms don't just weigh sources differently, they cite differently at a structural level. Some models pack a dozen URLs into a single answer; others cite sparingly, sometimes with no outbound link at all. Coverage rates for external links swing just as hard, with some platforms linking out in most responses and others rarely bothering. This isn't random. It reflects how each company built its retrieval and citation layer, and the effect on a pet brand is the same regardless of the reasoning behind it: visibility on one platform tells you nothing reliable about visibility on another.

Then there's the consensus problem. Ask the major models the same category question, and they agree on a top pick only a minority of the time. Full agreement across every model is rare enough to be worth noting when it happens. In practice, a supplement brand can hold strong visibility on Perplexity while sitting at near-zero on Gemini for the exact same query, run the same day.

Source weighting compounds it. A brand with strong veterinary endorsements and editorial placements gets rewarded on platforms that favor earned media. A brand with an active presence in pet owner forums gets rewarded where community sources carry weight. A brand producing video content with clean transcripts gets rewarded on platforms that index that kind of media. None of these advantages carry over automatically to a platform built around different sourcing habits.

Measuring AI visibility on one platform gives a false read. A brand's real exposure to pet owners is the sum across every platform those owners actually use, not the number attached to whichever tool a marketing team happened to check first.

What AI hallucinations mean specifically for pet brands

Hallucination, in plain terms, means an AI model states something wrong with total confidence: an ingredient that isn't in the formula, a price that changed two years ago, a health claim nobody ever made. In most consumer categories that's an annoyance. In pet care it's higher stakes, because a wrong ingredient list can matter enormously to an owner managing a dog's allergy, and a fabricated health claim can shape a decision about an animal's medical care.

Here's the odd twist: brands that get cited rarely are the ones most exposed to hallucination. A model with thin training data on a given brand has less real information to draw from, so when it's forced to answer anyway, it fills the gap with something that sounds plausible rather than something true. The brands most in need of accurate representation are the ones the model knows least about.

Waiting for an error to surface and correcting it after the fact is slower and far less reliable than making sure accurate, current, well-structured information exists for the model to draw from in the first place. That means brand-owned content with clear schema markup, vet-reviewed editorial, structured product data on retailer pages, and a real presence in the third-party reviews that feed these systems. Chasing hallucinations after they happen is a losing game. Building the source material that prevents them is the only version of this that scales.

The content signals that actually move AI citation share in the pet category

Researchers from Princeton, Georgia Tech, the Allen Institute for AI, and IIT Delhi published a study at ACM KDD in 2024 showing that adding citations, statistics, and quotations to existing content measurably improved how often that content got pulled into AI-generated answers. It's one of the few frameworks in this space that's actually been tested rather than assumed, and it gives pet brands a concrete starting point instead of a shrug.

A few habits follow from that research and from how these models parse content generally. Answer-first writing matters, since these systems extract direct answers, and content that buries the point three paragraphs down loses to content that states it up front. Complete, validated schema markup helps AI systems correctly read product attributes, ingredients, and claims instead of guessing at them. Vet-reviewed content signals the kind of sourcing AI platforms weight heavily in any health-adjacent category, and pet nutrition and pet health both qualify. Owned data on outcomes, feeding trial results, efficacy studies, gives a model something specific and citable instead of marketing copy it has no reason to trust.

None of this happens in a vacuum, separate from the wider information ecosystem. Getting placed in the outlets these platforms actually pull from (PetMD, The Spruce Pets, Kinship, Chewy's editorial content, the Amazon review ecosystem, veterinary trade press) does more for citation share than any amount of on-site optimization by itself. Platform-specific habits still apply on top of that: a community presence for forum-weighted platforms, video with real transcripts for platforms indexing visual content, long-form editorial for platforms that favor publication-style sourcing.

Brands that put structured content changes in place and pair them with real third-party amplification tend to see citation movement within weeks. This behaves more like a technical SEO fix than a rebrand.

How to measure AI share of voice in the pet category — and why one platform is never enough

Most pet brands can tell you their cost per click, their organic ranking for twenty keywords, and their Instagram engagement rate down to the decimal. Ask the same brand how often it shows up in AI answers compared to its competitors, and you usually get silence.

AI share of voice measures how often a brand appears in AI-generated responses across a representative set of category queries, compared against competitors, across platforms, not just one, platforms like AthenaHQ, for instance, track citation presence across eight or more LLMs simultaneously for exactly this reason. It behaves differently from every other share-of-voice metric marketers already track. There's no ad spend to control for, since none of this is paid; it reflects earned authority only. Platform variation means a single snapshot from one tool is close to meaningless on its own. And because the same query run twice on the same platform can return different results, a reliable read takes volume: many prompts, run repeatedly, not a handful of one-off checks.

Query design matters as much as platform coverage. A useful measurement plan spans pet food, treats, supplements, insurance, enrichment, and grooming, and cuts across life stages, breed types, and health conditions, not just branded searches for a company's own product names. 5WPR's Pet Industry AI Visibility Index takes this approach: it runs a large set of the prompts pet owners actually type, tracks which brands come up consistently across sub-categories, and flags the gap between where a brand sits in the market and where it sits in the citations.

Raw citation count isn't the whole story. Position matters, since being named first in an answer carries far more weight than showing up fifth on a list nobody reads that far into. Sentiment accuracy matters too, since a citation paired with wrong or outdated information isn't really a win. Consistency across repeated queries matters as well, since a brand that shows up once in ten tries has a very different problem than one showing up nine times in ten.

What a deliberate AI visibility program looks like for a pet brand today

The starting point is an audit: run the sub-category queries relevant to the brand, and map where citation share already exists, where it's missing entirely, and where the brand shows up with information that's simply wrong. That map becomes the priority list, not a guess about which category matters most.

From there, the gaps sort into three buckets, and each calls for a different response. Where a brand has real products but no citation presence (the way KONG's cat and bird toy lines lag behind its dog chew dominance), the fix is category-expansion content built for that specific gap. Where competitors hold citation share a brand should reasonably own given its market position, the fix is head-to-head content backed by vet-reviewed authority, not just more marketing copy. Where a brand shows up with wrong information already circulating, the fix is a structured correction: better schema, cleaner product pages, and placement in the third-party sources that actually feed these models.

The program has to run past the audit. Vet-reviewed content needs a real production cadence, not a single asset published once and forgotten. Internal data, trial results, efficacy studies, customer outcome summaries, needs to get surfaced in formats these models can actually cite. A brand also needs a sustained relationship with the third-party publications and platforms AI systems pull from, not a single press push timed to a launch.

Measurement has to become routine, tracked on the same cycle as paid search and SEO metrics, not checked once a quarter out of curiosity.

Fresh pet food is the clearest proof that none of this is fixed in stone. DTC leaders built early advantage the hard way, and now legacy giants are moving in with content footprints built over decades. Early citation share isn't permanent, but it doesn't correct itself either; it moves toward whoever puts in the structured work. Pet owners keep getting younger, the category keeps premiumizing, and more of that research starts with an AI assistant every quarter. The brands doing the structured work now are the ones that will still be in the answer a few years from here.