AI Visibility for Pet Brands: What Owners Ask and What AI Gets Wrong

Pet owners don't ask AI for "pet food". They ask about one animal, with one problem, often at the moment of purchase. Here are realistic examples, using placeholder brand names:

• "What's the best grain-free dog food for a senior Labrador with a sensitive stomach?"
• "Has Brand X dog food ever been recalled, and is it safe now?"
• "What's the first ingredient in Brand X salmon cat food?"
• "Does Brand X make a limited ingredient line for dogs with chicken allergies?"
• "Brand X vs Brand Y for a large-breed puppy: which has more protein?"
• "Which cat litter is low dust and safe for kittens?"
• "Where can I buy Brand X freeze-dried treats, and how much is a bag?"

Two patterns matter. Safety and ingredient questions are factual, so the engine either gets them right or wrong. Comparison and "best" questions are recommendation questions, where engines often lean on third-party pages. Self-description rarely moves those: see whether calling your product the best helps.

Also note that you can't see the real query. ChatGPT search rewrites a prompt into one or more targeted queries, and Google may fan out across subtopics. Treat your prompt list as a sample of owner intent, not the exact input.

Each engine draws on a different index, so measure each one separately.

• ChatGPT. Search answers depend on OAI-SearchBot. When search doesn't fire, the answer comes from training data with a knowledge cutoff.
• Google AI Overviews and AI Mode. Google says a page must be indexed and eligible to show with a snippet to appear as a supporting link, with no extra requirements. Classic Google SEO hygiene is the entry ticket.
• Microsoft Copilot. It grounds on Bing. In Otterly's 2026 test of two AI-generated sites, Copilot kept citing them after Google deindexed both, because Bing kept them. Your Bing index status is its own question.
• Perplexity and Claude. They have their own search crawlers (PerplexityBot, Claude-SearchBot) and their own robots.txt rules.

A brand can look healthy on one engine and absent on another. The step-by-step visibility check covers how to test them one at a time.

Pet brands carry a lot of specific, changing facts. That's exactly what models garble.

Recall news is the sensitive one. You can't edit what a model learned in training. You can make sure that, when an engine searches, it finds a dated, plain statement on your own site: what was recalled, when, which lots, and what has changed since. A specific fact on a page that can be fetched gives the engine a sentence it can quote.

The same discipline applies to ingredients. Put the full ingredient list, guaranteed analysis and life stage in plain text on each product page. Don't bury them in an image of the bag.

Many wrong pet answers are stale, not invented. The test separates the two.

1. Ask the same factual prompt with web search enabled and with it disabled. In ChatGPT, that means comparing a search-cited answer with one that shows no sources.
2. Note the dates. As of October 2026, OpenAI lists a Feb 16, 2026 knowledge cutoff for its GPT-5.6 models and Aug 31, 2025 for the Instant path. Anything you changed after those dates can't appear in an answer without search.
3. Compare. If the search-off answer is wrong and the search-on answer is right, the problem is the snapshot, and only a future model update will change it. If both are wrong, the engine is retrieving a bad page, and that you can fix.
4. When search is on, read the cited sources. They tell you which page to correct, yours or a retailer's.

If you just reformulated a recipe, run this test as part of a launch accuracy checklist.

For recommendation prompts, engines often cite pages other than yours. Being named correctly on those pages is associated with being in the answer, with no guaranteed effect. For pet brands, check:

• Online pet retailers and marketplaces. Ingredient panels, life stage, pack sizes and "discontinued" flags. Old variants left live are a common source of stale answers.
• Grocery and big-box listings if you sell there. The overlap with food and grocery brands is large.
• Review sites and owner forums. You can't edit them, but you can see which claims recur and correct the record on your own pages.
• News coverage of any recall. Know which articles engines cite, and make sure your own dated statement is easy to fetch.

Keep one internal fact sheet per product and match every listing to it. If your feeds go to many retailers, the ecommerce product data check walks through the fields that drift.

None of the content work matters if crawlers can't read the page. These checks have the strongest evidence behind them.

robots.txt, per bot. OpenAI states that each of its crawler settings is independent. GPTBot governs training. OAI-SearchBot governs ChatGPT search. A blanket AI block removes you from ChatGPT search without anyone noticing. A minimal allow for search crawlers looks like this:

User-agent: OAI-SearchBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: Claude-SearchBot
Allow: /

Firewall and CDN. A CDN can block a crawler that robots.txt allows, usually with a 403. From 2026-09-15, Cloudflare set Training and Agent crawlers to blocked by default on ad-displaying pages for new domains and existing free-tier customers, so access can change without you editing anything. The AI Crawler Checker requests a page as each AI crawler and shows which ones get turned away.

JavaScript rendering. Vercel's crawler study found ChatGPT and Claude crawlers fetch JavaScript files but don't execute them. Ingredient tabs, "nutrition" accordions and reviews loaded client-side may be invisible to them. Fetch the page with JavaScript off and check that the ingredient list is there.

Snippet controls. Check templates for noindex, nosnippet, max-snippet:0 and data-nosnippet. Each keeps content out of AI answers, and they're often inherited from a template by accident.

Say you run a hypothetical dog and cat food brand with three product lines. Here's a realistic month.

Days 1 to 7: baseline. Write 15 to 20 prompts across ingredients, safety, comparisons and where-to-buy. Run each on every engine you care about, with search on and off where you can. Log every wrong fact and the source it cites. The free checker on this site gives a quick snapshot for three buyer questions.

Days 8 to 14: access. Fix robots.txt per bot, test firewall responses, and make sure ingredient panels, life stage and pack sizes are in the raw HTML. Confirm index status in Google Search Console and Bing Webmaster Tools separately.

Days 15 to 21: facts. Publish plain statements on each product page: full ingredients, first ingredient, life stage, current sizes. Add a dated recall history page if you have one. Update retailer and marketplace listings to match, and retire discontinued variants.

Days 22 to 30: retest. Rerun the same prompts. Expect search-on answers to move first, and only where engines actually re-fetch. Search-off answers won't change until a model update. Track results per engine, never as one blended number. For the wider ecommerce picture, see AI visibility for ecommerce and DTC brands.

Should I add schema markup so AI quotes my ingredients correctly? There's no credible evidence that schema gets you cited in AI answers. Google says structured data isn't required for generative AI search, and an Ahrefs matched-control study found AI Overview citations fell 4.6% after pages added it. Existing markup does no harm. The thing that helps an engine quote your ingredients is the ingredient list itself, in plain text, on a page crawlers can fetch.

Can I get AI to stop mentioning an old recall? Not directly. If the recall is in a model's training data, answers given without search can repeat it until the model is updated. What you control is retrieval: a dated, factual page on your site explaining what was recalled, when, and what changed. When an engine searches, that page gives it an accurate sentence to use instead of an old article alone.

Does allowing GPTBot help my pet food show up in ChatGPT? No. OpenAI says GPTBot governs training use only, and its settings are independent of OAI-SearchBot, which controls whether your site can appear in ChatGPT search answers. You can block GPTBot and still allow OAI-SearchBot. Check both lines in your robots.txt, and then check that your firewall doesn't return a 403 to the search crawler.

Why does Copilot describe my products differently from Google's AI Overviews? They draw on different indexes and models. Copilot grounds on Bing, while AI Overviews require a page to be in Google's index with snippet eligibility. A product page healthy in one index can be missing or outdated in the other. Check Google Search Console and Bing Webmaster Tools separately, and test each engine on its own prompts rather than averaging them together.

Will publishing lots of pet care articles get my brand cited more? Volume is risky. In Otterly's 2026 experiment, two new sites with about 1,000 AI-generated posts each were removed from Google's index without warning. Google's spam policy targets many pages made to manipulate rankings without adding value. A few accurate, useful pages that answer real owner questions are a safer bet than scale.