30 Prompts Your Buyers Are Asking AI About Your Category, and Who Is Winning Them
Why Your Keyword Strategy Is Missing Half the Funnel
Your buyers have two research habits now. One looks like it always did: a short query in Google, a click to a listicle, a few tabs open. The other is newer and growing fast: a full question typed into ChatGPT or Perplexity, answered in a single response with brand recommendations already baked in.
According to 5W's 2026 First-Stop Study, 37% to 42% of consumers now begin brand research with AI tools instead of Google. Among Gen Z, 67% have used ChatGPT for brand research within the past six months. Adobe Digital Insights tracked retail AI-referred traffic during the 2025 holiday season and found it grew 693% year-over-year.
That's not a niche behavior. It's the mainstream purchase research pattern for a significant and growing share of your market.
Keywords are still important. But a keyword is a fragment. "Best protein powder" tells you someone has intent. A prompt is a complete thought: "what protein powder should I try if I'm lactose intolerant, train four days a week, and want to avoid artificial sweeteners" tells you who the person is, what they've already ruled out, and what they need to hear to buy.
The brands appearing in AI answers to those prompts aren't there because they bid on them. They're there because AI systems decided they were trustworthy sources, based on signals that have nothing to do with paid search.
This is the gap. Most brands have a keyword strategy. Very few have a prompt map.
What a Prompt Map Actually Is
A prompt map is a structured view of the natural-language questions your buyers ask AI at each stage of the purchase decision. The point isn't formatting. It's a framework for understanding what your buyer actually wants to know before they buy, which is a different question than what they type into a search box.
The difference matters because a prompt carries the full context of a buyer's decision. When someone types "best creatine for women" into Google, you know they're shopping. When someone types "I'm 43, just started lifting again after a few years off, and I've heard creatine helps with strength gains. Is it right for me and is it safe?" into ChatGPT, you know who they are, what they're afraid of, and what kind of answer will move them. A keyword doesn't get close to that level of signal.
A prompt map organizes those questions into three stages:
Awareness prompts: The buyer knows they have a problem or goal but doesn't know what category, ingredient, or brand might solve it. They're getting oriented.
Consideration prompts: The buyer knows the category. Now they're comparing brands, ingredients, formats, and specific fit for their situation.
Decision prompts: The buyer has a short list. They're asking AI to either confirm their choice or surface an alternative they haven't considered.
Each stage requires different kinds of presence. Awareness requires editorial coverage: the AI needs to have seen you mentioned as a credible category participant. Consideration requires review density and third-party comparisons. Decision requires being named by name in enough credible sources that the AI treats you as a safe recommendation.
Below is the full 30-prompt map for the health supplement DTC category, with real search volume data from Ahrefs (US, August 2026). These are the prompts your buyers are sending to AI right now.
Awareness Prompts (Prompts 1 to 10): The Buyer Is Getting Oriented
At the awareness stage, the buyer doesn't have a brand in mind. They're figuring out what the category is, whether it applies to their situation, and what the landscape looks like. AI responses here are category-level introductions, and the brands that appear are the ones with the broadest editorial footprint in health and fitness media.
| # | Prompt intent | Example prompt | Related keyword | Monthly US searches (Ahrefs, Aug 2026) |
|---|---|---|---|---|
| 1 | Category discovery | "what supplements help with energy and focus" | best greens powder | 13,000 |
| 2 | Problem/symptom | "why am I always tired even after sleeping" | n/a | Research intent |
| 3 | Format comparison | "powder vs capsules for creatine, what's the difference" | n/a | Research intent |
| 4 | Category overview | "what is creatine and should I take it" | best creatine | 42,000 |
| 5 | Beginner guidance | "where should a beginner start with supplements" | n/a | Beginner intent |
| 6 | Safety questions | "is it safe to take protein powder every day" | n/a | Trust intent |
| 7 | Gender-specific | "best supplements for women over 40 who just started exercising" | best creatine for women over 40 | 1,700 |
| 8 | Age-specific | "what electrolytes should someone in their 50s be taking" | best electrolyte drink | 20,000 |
| 9 | Goal-specific | "supplements for someone who wants to lose weight and keep muscle" | best protein powder for weight loss | 15,000 |
| 10 | Category validation | "is collagen actually worth taking or is it hype" | best collagen supplement | 3,500 |
Notice that most of the high-volume awareness searches ("best creatine": 42,000/mo; "best greens powder": 13,000/mo) are short-tail. But the prompts people actually type into ChatGPT are longer, more personal, and more specific. The keyword tracks intent. The prompt carries context a keyword never could.
A brand that shows up in response to "what is creatine and should I take it" is a brand the AI has decided is trustworthy enough to introduce the category. That's a very different kind of visibility than a Google ranking.
Consideration Prompts (Prompts 11 to 20): The Buyer Is Comparing
At the consideration stage, the buyer knows what they're looking for. They're comparing brands, digging into ingredients, and trying to figure out which option fits their specific situation. AI responses here are recommendation-heavy. They name brands, explain trade-offs, and synthesize third-party opinions.
| # | Prompt intent | Example prompt | Related keyword | Monthly US searches (Ahrefs, Aug 2026) |
|---|---|---|---|---|
| 11 | Product comparison | "LMNT vs Liquid IV for workouts, which is better" | best electrolyte drink | 20,000 |
| 12 | Ingredient deep dive | "what's the difference between creatine monohydrate and HCL" | best creatine monohydrate | 12,000 |
| 13 | Brand vetting | "is Thorne a reputable supplement company" | n/a | Trust intent |
| 14 | Side effects | "does creatine cause bloating or water retention in women" | n/a | Anxiety intent |
| 15 | Subcategory fit | "best creatine for women who don't want to look bulky" | best creatine for women | 31,000 |
| 16 | Quality signals | "what's the cleanest protein powder without heavy metals" | protein powder without heavy metals | 1,500 |
| 17 | Dosage guidance | "how much collagen should I take per day for joint pain" | n/a | Education intent |
| 18 | Review synthesis | "what do real people say about Vital Proteins collagen on Reddit" | n/a | Social proof intent |
| 19 | Value assessment | "is Ritual vitamins worth the price or is it overpriced" | n/a | Value intent |
| 20 | Stacking | "can I take creatine and protein powder together safely" | n/a | Optimization intent |
This stage is where the AI citation gap starts to matter most. Consideration prompts return answers that name specific brands. The brands named are the ones that appear most often in the editorial and community content the AI has indexed: review roundups, Reddit threads, practitioner recommendations, comparison articles.
A brand with a polished website and a large ad budget but thin third-party coverage will struggle to appear here. A brand reviewed by dozens of credible health publications and discussed across fitness communities will appear frequently, regardless of ad spend.
Decision Prompts (Prompts 21 to 30): The Buyer Is Ready to Purchase
Decision-stage prompts are the most valuable. The buyer is on the edge of purchasing. They're asking AI to confirm their choice, surface a better alternative, or resolve one final objection. A brand that appears at this stage in a favorable context is getting the last word before the purchase.
| # | Prompt intent | Example prompt | Related keyword | Monthly US searches (Ahrefs, Aug 2026) |
|---|---|---|---|---|
| 21 | Best-in-class | "what is the best creatine for men who lift heavy, 4x per week" | best creatine for men | 37,000 |
| 22 | Specific use case | "best protein powder for a woman trying to lose weight and tone up" | best protein powder for weight loss female | 2,900 |
| 23 | Subscription value | "is AG1 worth the monthly subscription or is there a better alternative" | n/a | Conversion intent |
| 24 | Where to buy | "where can I buy Thorne creatine at the best price" | n/a | Purchase intent |
| 25 | Discount / alternative | "is there a cheaper alternative to AG1 that's equally high quality" | n/a | Price sensitivity |
| 26 | Sample / trial | "does Transparent Labs offer samples before committing to a full order" | n/a | Commitment anxiety |
| 27 | Loyalty / repeat | "which creatine brand has the best subscription discount and free shipping" | n/a | Retention intent |
| 28 | Gifting | "best collagen supplement as a gift for my mom who has joint pain" | best collagen supplement for women | 2,100 |
| 29 | Bundle | "what protein and creatine stack should I buy for muscle gain" | best protein powder for muscle gain | 13,000 |
| 30 | Subscription vs one-time | "is it better to subscribe or just buy creatine one-time when I need it" | n/a | Commitment intent |
The "best creatine for men" cluster alone drives 37,000 monthly searches in Google. But people asking that via ChatGPT or Perplexity get a named recommendation, not a list of links to click through. The AI picks. If your brand isn't named, you don't get considered.
A Live Category Audit: Who's Winning in Supplements (August 2026)
We ran an AI citation analysis in Ahrefs for six supplement brands in August 2026. The metric is the number of times each brand's website is cited across ChatGPT, Google AI Overviews, Perplexity, and Gemini. These numbers come directly from Ahrefs site-explorer-ai-responses-count, pulled live for this piece.
| Brand | ChatGPT citations | Google AI Overviews citations | Perplexity citations | Gemini citations |
|---|---|---|---|---|
| Thorne (thorne.com) | 639 | 810 | 478 | 699 |
| Transparent Labs (transparentlabs.com) | 364 | 1,050 | 799 | 595 |
| Vital Proteins (vitalproteins.com) | 278 | 292 | 155 | 157 |
| Ritual (ritual.com) | 298 | 155 | 102 | 142 |
| LMNT (drinklmnt.com) | 213 | 194 | 295 | 144 |
| AG1 / Athletic Greens | ~0 | ~0 | ~1 | ~0 |
That last row deserves attention.
AG1 is one of the most aggressively marketed DTC supplement brands in the world. Podcast sponsorships, influencer deals, founder-led social content, homepage takeovers across fitness media. They've built enormous brand recognition through paid channels.
And their website is cited in approximately zero AI responses.
Meanwhile, Thorne (founded 1984, built on practitioner trust and clinical ingredient quality, running a fraction of AG1's media spend) appears in 639 ChatGPT citations and 699 Gemini citations.
This is the exact dynamic we've been writing about in this series. AG1 spent heavily on the channels they bought: podcast sponsorships, influencer deals, paid media placements. Those channels don't feed the AI layer. Thorne built editorial trust over four decades, and that's what the AI reads.
The divergence within the data is worth noting too. Transparent Labs leads on Google AI Overviews (1,050 citations) but trails Thorne on ChatGPT (364 vs 639). That gap reflects something specific: Transparent Labs has content structured and ranked in a way Google's own index rewards, but the broader editorial ecosystem that ChatGPT and Gemini pull from gives more weight to Thorne's clinical depth and practitioner reputation. The two platforms reward different inputs. Worth tracking both separately.
One note on methodology: the Ahrefs citation count reflects URL-level citations, specifically instances where an AI response links to or references a page from that domain. A brand can be mentioned in an AI response without generating a URL citation if the AI names the brand from training data rather than a retrieved source. These figures capture earned, source-backed citations specifically.
What's Driving the Gap
The difference between Thorne's 639 ChatGPT citations and AG1's near-zero isn't a mystery once you look at how AI systems actually build responses.
85% of brand mentions in AI answers come from third-party sources. AirOps and Kevin Indig's 2026 State of AI Search found that the vast majority of brand appearances in AI responses originate from content the brand didn't create: review sites, health publications, practitioner directories, comparison articles. If your content investment has been primarily in your own blog and ad campaigns, you're largely invisible to the AI's sourcing logic.
Community content punches above its weight. About 48% of Google AI Overview citations come from community platforms like Reddit and YouTube, according to the same research. A DTC supplement brand with an active customer community on Reddit (genuine discussions, real question-answering, transparent reviews) gets more AI representation than a brand with a premium editorial blog and no community footprint.
Content structure determines citability. Pages with sequential heading hierarchy and rich schema markup show 2.8x higher citation rates than unstructured pages (AirOps/Kevin Indig, 2026). The AI needs to be able to extract specific claims cleanly. If your product pages are JavaScript-heavy with no structured data, or if your key claims are buried in paragraph text, the AI often can't parse them confidently enough to cite them.
The opening of your content is where citations happen. SparkToro's analysis from January 2026 found that 44.2% of all LLM citations come from the first 30% of a piece of content (the introduction). The middle section accounts for 31.1% and the final 30% for just 24.7%. If your product page opens with brand storytelling and saves the substance for the third paragraph, you're losing the citation window before the AI gets to the part that matters.
Thorne's advantage isn't their current content strategy. It's that they've been publishing clinical content for over four decades, they're cited by practitioners and medical journals, and they have thousands of third-party editorial mentions that AI systems find credible and structurally parseable. That's a compounding asset. AG1 built their brand in a channel the AI doesn't read.
Why This Should Matter to Your Paid Ads Team
You might read the supplement table above and think: interesting academic exercise, but we run paid ads and this doesn't affect our acquisition numbers.
It does. Specifically, it affects the efficiency of every paid click you generate.
Adobe Digital Insights tracked AI-referred traffic during the 2025 holiday season and found it converted 31% better than non-AI traffic. Revenue per visit from AI referrals grew 254% year-over-year in that same dataset.
The reason isn't hard to explain. A buyer who clicked your ad after already asking ChatGPT about your category, and getting a favorable mention there, arrives with more intent and more trust than a buyer who clicked cold. They've had an intermediate step where an authoritative source told them you're credible. That buyer converts more readily.
Now flip it. A buyer sees your ad, clicks, lands on your product page, and then opens ChatGPT to check whether your brand is legit. If your brand doesn't appear in that AI response (or a competitor does), you just paid for a click that converted for someone else.
This is the mechanics behind why AI visibility and paid ads aren't separate strategies. They're part of the same funnel, and the AI layer is increasingly sitting in the middle of it.
We wrote about the ROAS side of this in A 4x ROAS Can Still Lose You Money. Here Is the Math. Optimizing for platform metrics without understanding the full funnel leaves money on the table. AI visibility is one of the hidden costs of ignoring it.
How to Run This Audit for Your Own Category
You don't need a Brand Radar subscription to start. The method below uses tools available to most brands and agencies.
Define your category precisely
Before you map prompts, you need a clear category definition. "Supplements" is too broad. "Creatine for women over 35" is a category you can audit. The more specific your category definition, the more useful your prompt map will be.
Build your 30-prompt map
Use the template in the next section. Replace the supplement-specific language with your product type, customer segments (gender, age, use case, life stage), and the specific objections your buyers have. The three-stage structure applies to every consumer category we've audited.
Check your AI citation baseline
In Ahrefs, run site-explorer-ai-responses-count for your domain (mode=subdomains). Do the same for your top two or three competitors. Record the citation counts per platform. This is your current share of AI presence, before you've done anything to improve it.
Audit your third-party coverage
Since 85% of brand mentions in AI come from third-party sources, the core question is: what does the editorial landscape around your brand look like vs. your competitors? Count the number of review sites, category roundups, and subreddit threads that mention each brand. The brand with the most diverse, credible third-party coverage will typically win the citation count.
Audit your content structure
For your top five pages (product detail, category, about, and any cornerstone blog content), check: Is there a clear H1 to H2 to H3 hierarchy? Are product attributes in structured lists? Is there schema markup (Product, Article, FAQPage)? Does the opening paragraph state the key claim clearly rather than building to it? Fix structure before adding volume.
Run the prompts manually
Open ChatGPT, Perplexity, and Google's AI search. Type the 10 highest-priority consideration-stage and decision-stage prompts for your category. Note which brands appear and how they're described. This qualitative read adds context the citation count doesn't capture. You see the actual language AI uses when recommending each brand, which tells you what it's pulling from and why.
The analysis takes half a day for someone who knows the category. The output is a priority list: editorial coverage gaps to close, content pages to restructure, review density to build. These aren't quick fixes, but they're the real drivers of AI visibility. Unlike paid channels, the gains compound.
The Full 30-Prompt Template
Use this table as your starting framework. Replace "supplement / creatine / protein powder" with your category and adapt the prompt language to match how your specific buyers talk.
| # | Buyer stage | Prompt intent | Template |
|---|---|---|---|
| 1 | Awareness | Category discovery | "what [category] helps with [problem or goal]" |
| 2 | Awareness | Problem/symptom | "why am I experiencing [symptom that your category solves]" |
| 3 | Awareness | Format comparison | "[format A] vs [format B] for [goal or outcome]" |
| 4 | Awareness | Category overview | "what is [ingredient or product type] and should I use it" |
| 5 | Awareness | Beginner guidance | "where should a beginner start with [category]" |
| 6 | Awareness | Safety questions | "is it safe to [use your product] every day" |
| 7 | Awareness | Gender-specific | "best [category] for [gender] over [age]" |
| 8 | Awareness | Age-specific | "what [category] should I use in my [decade or life stage]" |
| 9 | Awareness | Goal-specific | "[category] for someone who just [life event or situation]" |
| 10 | Awareness | Category validation | "is [product type] actually worth using or is it overrated" |
| 11 | Consideration | Product comparison | "[Brand A] vs [Brand B] for [specific use case]" |
| 12 | Consideration | Ingredient deep dive | "what's the difference between [variant A] and [variant B]" |
| 13 | Consideration | Brand vetting | "is [Brand] a reputable company" |
| 14 | Consideration | Side effects | "does [product] cause [specific concern]" |
| 15 | Consideration | Subcategory fit | "best [category] for [niche buyer segment]" |
| 16 | Consideration | Quality signals | "what's the cleanest [category] without [specific concern]" |
| 17 | Consideration | Dosage guidance | "how much [product] should I use per day for [goal]" |
| 18 | Consideration | Review synthesis | "what do real people say about [Brand] on Reddit" |
| 19 | Consideration | Value assessment | "is [Brand] worth the price" |
| 20 | Consideration | Stacking | "can I use [product A] and [product B] together" |
| 21 | Decision | Best-in-class | "what is the best [category] for [specific persona and context]" |
| 22 | Decision | Specific use case | "best [category] for [specific outcome] [qualifier]" |
| 23 | Decision | Subscription value | "is [Brand] subscription worth it" |
| 24 | Decision | Where to buy | "where can I buy [Brand] at the best price" |
| 25 | Decision | Discount / alternative | "is there a better alternative to [Brand] that costs less" |
| 26 | Decision | Sample / trial | "does [Brand] offer samples before I commit to a full purchase" |
| 27 | Decision | Loyalty / repeat | "which [category] brand has the best loyalty or refill program" |
| 28 | Decision | Gifting | "best [category] as a gift for [recipient and context]" |
| 29 | Decision | Bundle | "what [category] products should I combine for [specific goal]" |
| 30 | Decision | Subscription vs one-time | "is it better to subscribe or buy [product] one time" |
Running even the top 10 prompts manually takes less than an hour. You'll know more about your AI competitive position after that hour than you would from any keyword ranking report.
Frequently Asked Questions
How often do buyer prompts in a category actually change?
The core prompt structure is stable. The questions buyers ask about supplements, skincare, or fitness gear don't change dramatically quarter to quarter. What shifts is which brands appear in the answers, as editorial coverage evolves and new entrants build (or don't build) their AI presence. We recommend re-running your top 10 decision-stage prompts every quarter. Awareness and consideration prompts are worth a semi-annual check.
Does advertising spend affect my AI visibility?
Not directly, and the data is pretty clear on this. AI systems pull from indexed editorial content, third-party reviews, and community discussions. Ad impressions, paid landing page traffic, and influencer posts that aren't indexed as editorial content don't factor into it. The AG1 example above is the most concrete illustration we have: a brand spending tens of millions on media showing near-zero AI citations. Paid media builds awareness in the channels you're buying. Getting into the AI layer is a different project: it runs on editorial coverage, review density, and content AI can actually parse. The two tracks don't feed each other much.
How long does it take to start appearing in AI answers?
It depends on where you're starting from. Getting mentioned in a credible health or fitness publication can shift your AI citation count within weeks of that article being indexed. Building a review ecosystem (more product ratings, Reddit threads, YouTube unboxing videos) takes months of sustained effort. Content structure improvements (schema, heading hierarchy, cleaner opening paragraphs) can index within days, but structural changes only matter if you already have enough editorial authority to be in the AI's candidate set. From our observation, brands with 100 to 300 citations can move meaningfully within 60 to 90 days of targeted effort. Brands starting from near-zero need to address the third-party coverage gap first, which is a longer play.
What's the difference between Google AI Overviews citations and ChatGPT citations?
They use different source logic. Google AI Overviews favor sites that already rank well in Google's own index. Essentially Google amplifying its own signals. ChatGPT, Perplexity, and Gemini pull from a broader web index and weight editorial authority and recency differently from each other. The divergence in our supplement data makes this concrete: Transparent Labs leads on Google AI Overviews with 1,050 citations but trails Thorne on ChatGPT (364 vs 639). That gap suggests Transparent Labs has content that Google's crawler and algorithm reward, but the broader editorial ecosystem that ChatGPT draws from gives more weight to Thorne's clinical depth and practitioner reputation. A brand that wants full AI coverage needs to perform on both axes. The inputs for each are different.
How does this connect to what AI recommends for brand searches?
The citation count measures whether an AI response links to or cites your content. The brand mention is a related but separate metric. AI can name your brand from training data without linking to your site. The citation count is the more actionable signal because it reflects whether your content is specifically trusted enough to be sourced. You can explore the connection between mentions and citations further in We Asked ChatGPT to Recommend a Brand 100 Times. Here Is What Decided the Answer.
What to Do With This
The prompt map is only useful if you do something with it.
Start by running your top 10 decision-stage prompts today in ChatGPT, Perplexity, and Google AI search. Write down which brands appear and how they're described. Then pull your own AI citation count in Ahrefs and compare it to your two closest competitors.
That's your starting point. You'll know which stage of the funnel you're winning and which one you're losing. You'll know whether the gap is an editorial coverage problem, a review density problem, or a content structure problem. From there, you'll know what to fix first.
Want to see where your brand stands across both channels?
We audit AI visibility alongside organic and paid performance. Start with a conversation.
Talk to Us About Your CategoryLaunch into Success
Tell us a bit about yourself and your business. We are just one message away from the perfect partnership!