Your customers are increasingly asking ChatGPT, Claude, and Perplexity questions like “what’s the best tool for X?” instead of scrolling Google. When they do, one of two things happens: the model recommends you—or it recommends a competitor. Most brands have no idea which. This guide shows you how to find out with real buyer-intent prompts for D2C, SaaS, and hospitality, and how to turn a one-off check into ongoing AI citation tracking with the free SellOnLLM Chrome extension.
Why AI recommendations are the new shelf placement
When an answer engine names three products and yours isn’t one of them, you’ve lost the sale before the buyer ever visited a website. Unlike a Google results page, there’s no scrolling to position four—the model synthesizes a short answer and moves on. That makes “are we cited?” a revenue question, not a vanity metric. It’s the natural next step after you’ve fixed the fundamentals in our how to rank on ChatGPT guide and measured your AI Readiness Score.
Step 1: Write prompts the way buyers actually ask
The prompts you test have to mirror real purchase intent, not brand-name lookups. Testing “tell me about [your brand]” is useless—of course it knows you exist. What matters is the category question where you’re competing to be recommended. Focus on four shapes:
- Best-of: “best [category] for [audience]”
- Alternatives: “[competitor] alternatives”
- Head-to-head: “[competitor] vs [other]” (do you get mentioned as a third option?)
- Constraint-based: “affordable / fastest / [location] [category]”
Step 2: Buyer-intent prompts by industry
Here are starter prompt sets for the three verticals we build for. Adapt the brackets to your niche and geography.
D2C & ecommerce
- “best organic protein powder for women over 40”
- “affordable alternatives to [premium brand] skincare”
- “best sustainable running shoes for wide feet”
- “which coffee subscription has the freshest beans”
- “best gifts under $50 for a new dad”
For D2C, watch whether the model cites your product page, a marketplace listing, or a third-party roundup. If it’s always a roundup you’re not in, that’s your content gap. See the full D2C & ecommerce playbook.
SaaS & B2B
- “best CRM for a small marketing agency”
- “[competitor] alternatives for teams under 20 people”
- “cheapest project management tool with time tracking”
- “best AI note-taker that integrates with Slack”
- “[competitor A] vs [competitor B] for onboarding speed”
SaaS buyers lean heavily on “alternatives” and “vs” prompts—exactly the query shape covered in our comparison content pillar. More in the SaaS & B2B guide.
Hospitality & hotels
- “best boutique hotels in [city] for couples”
- “pet-friendly hotels near [landmark]”
- “where to stay in [city] with free parking and a pool”
- “best hotels in [neighborhood] under $200 a night”
- “family-friendly resorts in [region] with a kids club”
Hospitality answers lean on local entities and reviews, so consistency across your site and profiles is critical. See the hospitality vertical.
Step 3: Read the answer like an analyst
Running the prompt is only half the job. For each response, record:
- Are you mentioned at all? Presence vs absence is the first binary.
- Position and framing — first pick, honorable mention, or a caveat (“some people also use…”).
- Who is cited instead — the competitors winning the recommendation.
- What source it pulled from — your site, a review platform, or a third-party list you could get into.
Do this manually once and you’ll immediately see the pattern: models cite whoever is clearest, most structured, and most consistently mentioned across the web. That’s the same logic behind the trust signals in our E-E-A-T scanner.
Step 4: Turn spot-checks into monitoring
Manual testing has three problems: answers are non-deterministic (they vary run to run), you can’t easily cover three engines at once, and you won’t remember last month’s baseline. That’s where the extension’s AI Visibility feature comes in. It:
- Runs your buyer-intent prompts across ChatGPT, Claude, and Perplexity.
- Records whether you’re cited, your position, and which competitors appear.
- Tracks the trend over time so you can prove your AEO work is moving the needle.
- Suggests prompts you might not have thought to test.
You get one free AI Visibility check with the extension, then monitoring on the paid tier. The point is to make citation tracking a repeatable dashboard, not a one-off curiosity. This mirrors how our AI Visibility MCP works for Claude users.
Step 5: Close the gap when you’re not cited
When a prompt shows a competitor instead of you, the fix usually falls into one of these buckets:
- Clarity — your page doesn’t plainly say who it’s for. Add a “best for” summary near the top.
- Structure — missing FAQ or comparison content that models love to lift. Add FAQ schema.
- Coverage — you’re absent from the third-party roundups the model trusts. Earn inclusion in those lists.
- Consistency — your brand facts differ across the web, so the model can’t build a confident entity. Align them.
Fix the page, re-run the audit for a fresh AI Readiness Score, then re-run the prompt a few weeks later to confirm movement. For the strategy layer, the AEO Hub and our why track AI brand visibility post go deeper.
A note on how AI engines pick sources
Answer engines increasingly cite the live web. Perplexity documents its crawling behavior in its bots guide, and OpenAI describes how its crawlers access pages. If those crawlers can’t read your content—or you’ve blocked them in robots.txt—you can’t be cited no matter how good your copy is. We break this down in the AI crawlers pillar.
Start with one prompt today
Pick the single most valuable “best [category] for [audience]” prompt in your business and ask all three engines right now. If you’re not the answer, install the extension, run the AI Visibility check, and start closing the gap. It’s the fastest way to find out whether the AI shelf has your brand on it.