AI Visibility for Beverage and Spirits Brands

Beverage buying questions fall into three groups: what to drink, what a bottle is, and where to get it. The second and third groups are where brands lose, because they depend on precise facts.

Realistic example prompts:

• "What's a smooth bourbon under $50 to give as a gift?"
• "Which mezcal works best in a mezcal negroni?"
• "What's the ABV of the cask strength release from [brand]?"
• "Is [brand] tequila additive-free?"
• "Does [brand] still make its peated expression?"
• "Where can I buy [brand] gin near me?"
• "Best non-alcoholic spirits for a dry month"
• "Which hard seltzer has the least sugar?"

Recommendation prompts reward being named on the sites engines retrieve. Fact prompts reward having the exact number on a page an engine can fetch. Where-to-buy prompts lean heavily on retailer and local listings you don't own.

Say you run a hypothetical craft distillery with four expressions. A buyer asks for a gift bourbon and gets three competitors. Then they ask about your rye and get last year's proof. Those are two different problems with two different fixes.

Treat each engine as its own channel. They use different indexes, different crawlers and different models.

• ChatGPT searches at the model's discretion. OpenAI's help page says ChatGPT search rewrites the user's question into one or more targeted queries, so the exact prompt you test is not the query it runs.
• Google AI Overviews and AI Mode draw on Google's index. Google says a page must be indexed and eligible to show with a snippet to be a supporting link, with no extra technical requirements.
• Gemini is a separate product from AI Overviews, governed partly by the Google-Extended control.
• Perplexity runs its own index via PerplexityBot and cites sources inline, which suits where-to-buy questions.
• Claude decides when to search, and on claude.ai search is a user toggle. See what the evidence says about Claude visibility.
• Microsoft Copilot grounds on Bing's index, so Bing coverage is its own check.

When search is off, an engine answers from training data frozen at a cutoff. OpenAI's model docs list a February 16, 2026 cutoff for the GPT-5.6 family and August 31, 2025 for the Instant path. A release you launched after those dates simply does not exist in those answers. Nobody can change that this quarter.

Drinks facts are unusually hard for AI to get right. Every reason is checkable.

• ABV and proof vary. The same expression can ship at different strengths by market or batch. If your site gives one number and a retailer gives another, the engine picks one.
• Product lines churn. Limited releases, seasonal editions and discontinued bottles linger on retailer pages and in training data long after they're gone.
• Availability is local and regulated. Distribution differs by state and country. An engine that finds one retailer listing may say a bottle is sold everywhere, or nowhere.
• Spec sheets hide in scripts. Tasting notes, ABV and age statements often load through tabs, carousels or product finders that only render after JavaScript runs.
• Ownership changes. Acquisitions and brand sales leave old parent-company facts in circulation.

The one on-page move with positive evidence is plain: state each specific fact once, clearly, on a page that can be fetched. Give each expression its own page with the ABV, size, age statement and markets in the HTML. That gives an engine a sentence it can quote. It does not guarantee the engine uses it.

Age gates are the access problem most specific to this industry. How the gate is built decides what a crawler receives.

Vercel's crawler study found ChatGPT and Claude crawlers fetch JavaScript files (11.50% and 23.84% of their requests) but don't execute them, so they can't read client-side rendered content. Googlebot renders JavaScript. Most AI crawlers don't.

That creates three common outcomes:

1. Gate as an overlay, content in the HTML. The crawler likely receives the product facts underneath.
2. Gate that loads content only after the visitor confirms. The crawler receives the gate and nothing else, on every page.
3. Gate as a server redirect. Visitors without a confirmation cookie, including crawlers, may be sent to the gate page instead of the product.

Test it directly. Fetch a product URL with JavaScript disabled and no cookies, then compare the text with what a browser shows after passing the gate. If the ABV, expression names and descriptions are missing from the raw HTML, AI crawlers are likely missing them too. Also check whether the gate page carries noindex or nosnippet, which would keep it out of AI answers entirely.

How the gate works is a compliance decision first. Whatever design you choose, serve crawlers and people the same content. Showing AI bots a different page from visitors is cloaking. It is a policy risk, not a fix.

After the age gate, run the standard access checks. Each one removes a block. None adds a ranking boost.

robots.txt, per bot. OpenAI states its crawler settings are independent. GPTBot governs training use. OAI-SearchBot governs whether you appear in ChatGPT search answers. A blanket AI block often removes search eligibility by accident. A labelled example that opts out of training but stays eligible for ChatGPT search:

User-agent: GPTBot
Disallow: /

User-agent: OAI-SearchBot
Allow: /

Check PerplexityBot, Claude-SearchBot, Googlebot and bingbot the same way. Some answer-time fetchers behave differently: Perplexity says Perplexity-User generally ignores robots.txt. The difference is explained in index-time crawlers vs answer-time fetchers.

Firewall and CDN. A firewall can return a 403 to a bot that robots.txt allows. OpenAI's help page says eligibility needs both crawler permission and a host or CDN that accepts its published search bot IP addresses. As of October 2026, Cloudflare blocks Training and Agent crawler categories by default on ad-displaying pages for new domains and existing free-tier customers, a change that took effect on September 15, 2026. Read robots.txt vs firewall for both gates side by side. The AI Crawler Checker requests a page as each AI crawler and shows which ones get turned away.

Index status. Check Google Search Console and Bing Webmaster Tools separately. Healthy in one tells you nothing about the other.

When engines cite other sites instead of yours, being named on those sites is the route into the answer. The evidence shows a strong association, not a guaranteed effect. For drinks, audit these first:

• Retailer product pages. National chains, online bottle shops and delivery apps. Check ABV, size, price and whether discontinued bottles still show as available.
• Distributor and retailer store finders. These feed where-to-buy answers. Stale stock data becomes a wrong answer.
• Ratings and review sites. Spirits rating databases and competition results pages often carry the expression names and proofs engines repeat.
• Cocktail and recipe sites. These drive recommendation prompts for mixing spirits.
• Forums. Enthusiast communities and Reddit threads about your brand shape recommendation answers.
• Reference sites. See whether you need a Wikipedia page before chasing one.

Send corrections to retailers with the right figure and a link to your own product page. The same logic applies to any brand sold through resellers, covered in AI visibility for ecommerce and DTC brands.

Week 1: baseline

Write 10 to 15 prompts across recommendation, fact and where-to-buy questions. Run them on each engine and record who is named and which facts are wrong. The free checker on this site shows a quick snapshot for three buyer questions.

Week 2: access

Test the age gate with JavaScript off and no cookies. Check robots.txt per bot, firewall responses per crawler, and noindex or nosnippet on gate and product pages. Confirm index status in Google and Bing.

Week 3: your own facts

Give every current expression its own page with ABV, size, age statement and markets in the raw HTML. Mark discontinued bottles as discontinued instead of deleting them silently. Remove contradictions between product pages, the about page and press materials.

Week 4: the wider web

Send corrections to your top retailers and rating sites. Update store finder data. Re-run the week 1 prompts and compare per engine.

Expect some answers to stay wrong. Engines decide what to retrieve and cite, and answers without search still reflect old training data. Your job is to remove every block you control and make the right fact easy to find.

Should a spirits brand block GPTBot? That is a training decision, not a visibility one. OpenAI states GPTBot governs whether content may be used for training, while OAI-SearchBot controls appearing in ChatGPT search answers, and each setting is independent. You can block GPTBot and still allow OAI-SearchBot. Nobody has measured whether training exclusion changes how a model recalls your brand, so treat it as a policy choice rather than a ranking lever.

Do I have to remove my age gate to show up in AI answers? No. The question is what the gate hides. If your product facts are in the HTML a crawler receives, a gate can sit on top of them. If content loads only after a visitor confirms their age, crawlers that don't run JavaScript likely see only the gate. Test the raw HTML, and agree any change with your compliance team. Never serve bots a different page from people.

Can I pay to get my bottle recommended in ChatGPT? Organic recommendations are not something you buy placement in. OpenAI describes ranking with "multiple factors" and says placement is not guaranteed. What you control is access, accurate facts on fetchable pages, and correct listings on the sites engines cite. More detail is in [whether you can pay to appear in ChatGPT](/questions/pay-to-appear-in-chatgpt).

Will a new model release fix wrong ABV in AI answers? Not reliably. A new model brings a new training cutoff, so answers without search may update, or may keep repeating whatever outdated figure circulated widely. Answers with search depend on what pages the engine retrieves on the day. Fixing the figure on your own pages and on retailer listings is the part you control. See [how model releases change brand answers](/questions/new-ai-model-changes-brand-answers).

Does adding product schema help AI engines quote my ABV? The evidence does not support schema as a lever for AI answers. Google says structured data isn't required for its generative AI features, and an Ahrefs matched test found AI Overview citations fell 4.6% after pages added schema. Existing markup does no harm. The stronger move is stating the ABV in plain text on a fetchable product page.