1. TL;DR

Today's wellness buyers open an AI assistant first and reach your website later, if the assistant sends them. The assistant assembles each answer from the third-party sources it trusts, so on-site SEO alone leaves your brand outside the answer. Earned media drives 84% of AI citations (Muck Rack, 2026), while a brand's own site supplies 5 to 10% of what AI search references (McKinsey, 2025). A brand missing from those sources misses its highest-converting buyers. Wellness sits at the center of the shift: beauty, personal care, and consumer health absorb the largest share of AI shopping referrals (Euromonitor, 2026). The shoppers arriving through AI carry intent: AI-referred retail traffic grew 138% year over year in May 2026 and converted 54% better than other traffic (Adobe Analytics, 2026). Winning a place in those answers is a discipline we call Selection Engineering™: earning your brand's place in the third-party sources AI cites, then confirming those sources describe you right.

STAT BLOCK
- Consumers using AI for shopping help, past 6 months: 77.6% (Semrush / Exploding Topics via Search Engine Land, May 2026)
- AI-referred retail traffic growth, year over year: +138% (Adobe Analytics, May 2026)
- AI-referred retail traffic growth since Oct 2024: +1,324% (Adobe Analytics, 2026)
- AI-referred traffic conversion vs non-AI: +54% (Adobe Analytics, May 2026)
- Share of AI citations from earned media: 84% (Muck Rack Generative Pulse, May 2026)
- Brand's own site as share of AI-search sources: 5 to 10% (McKinsey, October 2025)
- Generative AI shopping referrals, beauty + personal care: ~45% (Euromonitor, 2026)
- Generative AI shopping referrals, consumer health: ~28% (Euromonitor, 2026)
- US consumers who call wellness a top or important priority: 82% (McKinsey, 2024)
- Global wellness economy, 2024: $6.8T, forecast $9.8T by 2029 (Global Wellness Institute, November 2025)
- AI users who verify recommendations before buying: 86% (Product.ai, April 2026)

2. The Shift: How Wellness Buyers Decide in 2026

AI has become a standing habit in the buying journey. Semrush and its Exploding Topics team surveyed 1,009 US consumers and found 77.6% used AI to help with shopping or purchasing decisions in the past six months (Search Engine Land, 2026). Even among consumers over 60, more than half reported the same behavior. A shopper reaches for ChatGPT the way that shopper once opened a browser tab.

The buyer often researches through AI, then confirms the answer somewhere else. ChannelEngine's Marketplace Shopping Behavior Report 2026, built on a survey of 4,500 marketplace shoppers, found 58% used AI tools to research products and 37% started a purchase journey through an AI assistant (ChannelEngine, 2026). Product.ai's Trust in AI Commerce Report, run in April 2026 against 1,463 US online shoppers, found that among the 43% who used AI for product research, 86% verified the recommendation through another source before buying (Product.ai, 2026). For a health or wellness purchase, that verification step matters more, because a buyer weighs the risk to the body.

AI traffic now buys, and it buys at a higher rate. Adobe Analytics, drawing on more than one trillion visits to US retail sites, reported that AI-referred retail traffic grew 138% year over year in May 2026 and 1,324% since October 2024 (Digital Commerce 360, 2026). Those visitors converted 54% better than non-AI traffic and generated 53% more revenue per visit, a reversal from a year earlier when AI visitors converted at close to half the rate (Adobe, 2026). The click from an AI answer arrives late in the decision, so it carries more intent.

Wellness sits at the center of the shift. Euromonitor International's panel of 5.3 million AI users found beauty and personal care captured about 45% of generative AI shopping referrals and consumer health about 28% (Forbes, 2026). Euromonitor tied the concentration to categories where buyers face dense claims and need help comparing options, which describes supplements, recovery devices, and sleep tech with precision. Euromonitor's press release (Business Wire, March 18, 2026) reported AI referrals to retail websites growing 304% from January to December 2025, against 40% growth from other referral sources, with ChatGPT referrals to US websites up 367% in 2025.

Wellness holds its budget priority, and buyers build stacks. McKinsey's Future of Wellness research found 82% of US consumers call wellness a top or important priority in their daily lives (McKinsey, 2024), and Circana measured $1.1 trillion in US consumer spending power directed at well-being (Circana via Fitt Insider, 2025). Fitt Insider's 2026 Halftime Report describes wellness buyers assembling stacks of products and routines across sleep, recovery, and supplementation (Fitt Insider, 2026). A buyer assembling a stack asks comparison questions, and comparison questions are where AI referrals concentrate.

The category itself keeps expanding. The Global Wellness Institute measured the wellness economy at $6.8 trillion in 2024, growing 7.9% from the prior year, and forecast $9.8 trillion by 2029 (GWI, 2025). Sub-categories inside the report track the products wellness founders sell: infrared saunas move toward $3.64 billion by 2033 (Coherent Market Insights, 2026), and red light therapy grows from about $587 million in 2026 toward $1.13 billion by 2033 (Grand View Research, 2026). A growing market and a new discovery channel arrive together, which raises the stakes on visibility.

Half of consumers treat AI search as a primary source. McKinsey's AI Discovery Survey, fielded in August 2025 among 1,927 US consumers, found half intentionally seek out AI-powered search, and 44% of those users call it their primary and preferred source of insight, ahead of traditional search at 31% (McKinsey, 2025). McKinsey projects $750 billion in US consumer spend will flow through AI-powered search by 2028 and estimates 20 to 50% of traditional search traffic sits at risk (McKinsey, 2025).

3. Where AI Answers Actually Come From

AI answers come from third-party sources, not from your website. When a buyer asks an AI engine for a product recommendation, the engine assembles the answer from the sources it has read and trusts. Muck Rack's Generative Pulse study, based on analysis of more than 25 million cited links across ChatGPT, Claude, and Gemini, found earned media accounts for 84% of AI citations, journalism alone for 27%, and paid or advertorial content for 0.3% (Muck Rack, 2026). The figure has held between 82% and 89% across three editions since July 2025.

Your own site is a minority of what AI reads. McKinsey found a brand's own site comprises 5 to 10% of the sources AI search references, with the remainder coming from publishers, affiliates, and user-generated content (McKinsey, 2025). In consumer-packaged goods and financial services, more than 65% of AI-search sources are publishers, user-generated content, and affiliate sites (McKinsey, 2025). Supplement brands face the same pattern: 5W's Supplements AI Visibility Index 2026 estimated brand-owned pages supply about 5 to 8% of observed supplement citations, while editorial health media supply 40 to 45% and specialist research sources such as Examine.com, ConsumerLab, and PubMed-indexed studies supply 15 to 20% (5W, 2026).

Citation share concentrates fast inside a category. The same 5W index, built on 3,800 supplement prompts run from January through April 2026, found four brands, Thorne, Seed, AG1, and Momentous, together captured more than 47% of observed supplement citation share (5W, 2026). Four brands hold close to half the answer space, and the brands behind them start near zero. Concentration of that kind rewards the brands that treat AI visibility as a program and punishes the brands that wait.

"Selection" is the moment an AI engine chooses which sources to build its answer from. An engine reads the web, ranks sources by trust and relevance, and cites a handful. A brand does not control that selection. A brand influences it by existing, described right, inside the sources the engine already reads.

Selection Engineering™ is Prominently AI's framework for earning a brand's place in the third-party sources AI engines cite. The practice works in two moves. First, earn presence in the earned media, review platforms, and community threads AI reads. Second, confirm those sources describe the brand right. In wellness, the earned layer has names: expert-driven media such as Wellworthy and Fitt Insider, category review platforms, specialist research sources, and the threads where buyers compare their stacks. Expert roundups, awards, and vetted directories sit inside the earned layer that drives 84% of citations (Muck Rack, 2026). Selection Engineering treats the AI answer as an editorial ecosystem a brand influences through evidence, rather than a page a brand controls. For a wellness brand under FDA and FTC claim rules, earned editorial coverage carries a second benefit: it builds AI visibility and holds up under regulatory review at the same time.

4. Per-Platform Notes

The three engines Prominently tracks each pick sources on different logic. A brand can hold a strong position in one and stay absent from another. Citation patterns shift within weeks when platforms change models or lose data access, so every figure below carries its measurement source and date, and the October 2025 Reddit-Perplexity split shows how fast the ground moves.

ChatGPT. OpenAI first cited the 900 million weekly active user mark on February 27, 2026, up from 800 million in October 2025 and double the 400 million of a year earlier (Search Engine Land / TechCrunch, 2026). ChatGPT builds answers from a mix of training knowledge and selective web retrieval, and it leans toward consensus, encyclopedic sources: Wikipedia is its single most-cited domain, accounting for close to 48% of its top-10 source share (5W State of AI Citations, June 2026). ChatGPT cites fewer sources per answer than its rivals (Discovered Labs, 2025), which sets a higher bar for entry. Its Shopping Research feature surfaces three to eight products from structured product feeds and third-party trust signals. OpenAI's Help Center states that product results are organic and unsponsored, ranked on relevance, with merchants ranked "based on factors like availability, price, quality, and whether they are the maker or primary seller of that item" (OpenAI, 2026). Moving ChatGPT takes sustained third-party presence and clean, consistent brand description across the web.

Perplexity. Perplexity runs a live web search on each query and cites sources inline, and it cites more sources per answer than ChatGPT (Discovered Labs, 2025), so more slots open for smaller brands. Its user base sits near 34 million monthly active users (SimilarWeb via Business of Apps, March 2026), a fraction of ChatGPT's reach, with an audience skewed toward research-heavy queries. Its sourcing shifted hard in late 2025: Reddit supplied 46.7% of Perplexity's top citation sources through mid-2025 (Profound, 2025), then Reddit sued Perplexity in October 2025 (Reuters, 2025), and Reddit's share of Perplexity citations fell from 25% in February 2026 to 7% in April while YouTube filled the gap (Tinuiti via MediaPost, 2026). Perplexity favors fresh content, and year signals such as "2026" in titles lift citation rates by about 30% (Leapd, 2026). A wellness brand present in current editorial coverage and cited video content has a real path into Perplexity answers.

Google AI Overviews. AI Overviews reach more than 2 billion monthly users (Google, 2026) and trigger on approximately 48% of tracked queries as of February 2026, a 58% year-over-year increase (BrightEdge Generative Parser via Search Engine Journal, 2026). They draw from Google's index enriched with reinforced E-E-A-T signals and a re-ranking system built for AI queries. Strong traditional rankings give a head start, though ranking position no longer guarantees a citation: only about 12% of AI citations overlap with Google's organic top 10 (Discovered Labs, 2025). Reddit and YouTube feature often, with Reddit near 21% of AI Overview citations (5W / Profound, 2026).

Concentration across all three. 5W's AI Platform Citation Source Index 2026, a synthesis of more than 680 million citations across the major engines, found the top 15 domains capture 68% of consolidated AI citation share, with Reddit the single most-cited domain at close to 40% frequency across engines (5W Public Relations via PR Newswire, 2026). 5W produces this index to market its own services, so read it as a directional synthesis rather than peer-reviewed measurement. The takeaway holds across independent studies: a small set of trusted domains decides most answers, and the roster of those domains changes faster than a quarterly plan.

5. Build Your Prompt Panel

Section 4 shows each engine reads a different slice of the web, so the way to learn where your brand stands is to ask the engines what your buyers ask. You could borrow someone else's prompt list, and it would test someone else's market. A panel built from your own buyers, rivals, and doubts tests yours. The method takes four templates, five rules, and fifteen prompts.

The four templates

Buyers move toward a purchase in four stages, and each stage has a prompt shape. Fill the brackets and you have a prompt.

Discovery. The buyer knows the problem and wants candidates.
best [product type] for [person, outcome, or constraint]

Comparison. Two names in the ring.
[rival A] vs [rival B]
Run the self-audit version too: [your brand] vs [the rival you lose deals to]

Trust. One brand under inspection.
is [brand] legit · does [product] work or is it hype

Purchase timing. Money in hand, one doubt left.
is [product] worth the money · when does [brand] go on sale

Five rules for sounding like a buyer

1. Leave the doubt in. Buyers type "does it work or is it hype," and a prompt with the doubt kept in returns a different answer than a clean one. Your panel should ask the question the way your buyer asks it, suspicion included.
2. Add one constraint. "For hot sleepers." "Under $500." "For a small apartment." Buyers narrow with constraints, and a constrained prompt shows whether AI names you in the niches you win.
3. Name real rivals. The comparison stage runs on names. Prompt the matchup you lose deals to, and prompt the matchup you win, so the panel covers both directions.
4. Skip the polish. Lowercase, missing punctuation, run-on phrasing. Engines parse intent, and buyers type fast. A polished prompt tests a question nobody asks.
5. Date the timing prompts. Add the year: "worth it in 2026." Buyers add the year on purchase questions, and year signals lift citation rates on Perplexity by about 30% (Leapd, 2026).

The panel method

Fifteen prompts make a working panel: five discovery, four comparison, three trust, three timing. Your brand goes in the trust and timing slots, your rivals go in the comparisons, and the category carries discovery. Run each prompt in ChatGPT, in Perplexity, and in Google with AI Overviews on, then log each result using the four states in the next section: named, cited, absent, or wrong. Date the log.

One run is a snapshot, so rerun the same fifteen in 30 days and compare. AI answers move, and the movement is the signal. Fifteen prompts across three engines takes most founders under an hour, and the hour tells you whether AI sends buyers toward you, past you, or away from you.

A worked example

Say you sell a vagus nerve stimulation device. Your panel, built from the templates:

Discovery
1. best vagus nerve stimulation device for anxiety
2. how to stimulate the vagus nerve at home devices
3. best wearable for stress and HRV
4. devices that calm your nervous system that aren't apps
5. vagus nerve stimulator for sleep which one

Comparison
6. [your brand] vs Pulsetto which works better
7. Pulsetto vs Apollo Neuro which works better
8. Sensate vs Apollo Neuro for anxiety
9. vagus nerve device vs meditation app results

Trust
10. is [your brand] legit does it actually work
11. [your brand] reviews reddit
12. are vagus nerve stimulation devices science or a scam

Purchase timing
13. is [your brand] worth the money
14. is a vagus nerve device worth it in 2026
15. [your brand] discount code or wait for a sale

Swap the brackets, run the fifteen, log the states. The panel measures your market from the first run.

6. Reading Your Results: The Four-State Scorecard

Run a prompt through an engine, and your brand lands in one of four states. Each state points to a different action.

Named. The engine names your brand in the answer and frames it as a recommendation. You hold the position buyers act on. The work here is defense: keep the earned coverage and reviews behind that answer fresh, because citation share shifts within weeks, not years.

Cited. The engine links a source that mentions your brand without recommending it, or lists you among alternatives. You sit inside the ecosystem, and you have a foundation to build on. The work is to deepen presence in the sources that carry weight for your category, so a mention grows into a recommendation.

Absent. The engine answers the category question and never mentions your brand. Your competitors own the answer, and your buyers never see you at the moment of decision. The work is to earn presence in the third-party sources the engine reads for that prompt, because on-site content alone will not close the gap.

Wrong. The engine names your brand and describes it with a stale price, a discontinued product, a mistaken claim, or a competitor's feature. This state carries the highest risk for a wellness brand, because an inaccurate health claim in an AI answer can mislead a buyer and expose you to scrutiny. The work is correction at the source: fix the third-party pages and structured data the engine drew from.

A single prompt is a snapshot, not a score. Track the four states across many prompts and all three engines over time to see the real pattern.

7. What to Do About It

Three disciplines cover the work. A brand needs each one, and they build on each other.

SEO: Get Found. Clean, crawlable, well-structured pages remain the foundation. AI engines read your site to confirm facts, and Adobe found many retail product pages stay less readable to AI systems than other page types (Adobe, 2026). Structured product data, clear headings, and stable URLs let an engine extract and trust what you publish. This discipline gets your brand found.

AI Search Optimization: Get Cited. Since earned media drives 84% of citations (Muck Rack, 2026) and your own site supplies 5 to 10% of sources (McKinsey, 2025), the citation work happens off your domain. Earned editorial coverage, credible review presence, and participation in the community threads AI reads move a brand from absent to cited to named. In supplements, four brands hold 47% of the citation share (5W, 2026), which shows the ceiling for the brands that run this work and the gap facing the brands that skip it. This discipline gets your brand cited.

Brand Reputation: Stay Accurate. AI engines index positive and negative mentions at close to equal rates, so an inaccurate or outdated source can surface inside an answer. For a category under FDA and FTC claim rules, accuracy is both a trust matter and a compliance matter. Monitoring what the sources say about your brand, then correcting errors at the source, keeps AI answers accurate. This discipline keeps your brand accurate.

Prominently runs the three disciplines under the names SEO, AI Search Optimization, and Brand Reputation, with daily prompt tracking across ChatGPT, Perplexity, and Google AI Overviews. The point of naming them is clarity: each discipline does one job, and a brand can see which job it needs first.

8. The Turn

You now know where AI answers come from and how to read your own results. The next step is to see your four states for a set of real buyer prompts. Prominently's Visibility Scanner runs your brand across ChatGPT, Perplexity, and Google AI Overviews and returns where you are named, cited, absent, or wrong. Run your own scan on prominently.ai and start from data about your brand, in your category, today.

9. Sources Appendix

Each entry flags its publication date. Recency matters, and 2026 sources carry more weight than 2025 sources where both exist.

1. Search Engine Land, "New data: 77% use AI to shop. Nearly 1 in 3 won't let it spend" (Semrush / Exploding Topics survey, n=1,009). Published May 2026.
2. Digital Commerce 360, "Adobe: AI-referred traffic to retail sites doubles in a year" (Adobe Analytics, May 2026 data). Published June 17, 2026.
3. Adobe Business, "AI traffic grows but retail sites lag in AI search visibility" (May 2026 data). Published 2026.
4. Forbes (Michelle Evans), "Why AI Will Reshape Beauty More Than Grocery Shopping" (Euromonitor Passport E-Commerce, panel of 5.3M AI users). Published July 9, 2026.
5. Euromonitor International / Business Wire, "AI-driven Referrals Grew Over 300% in 2025." Published March 18, 2026.
6. Global Wellness Institute, "The Global Wellness Economy Hits a Record $6.8 Trillion and Is Forecast to Reach $9.8 Trillion by 2029." Published November 19, 2025.
7. Muck Rack, "Earned media still drives 84% of AI citations" (Generative Pulse, 25M+ links). Published May 7, 2026.
8. McKinsey & Company, "New front door to the internet: Winning in the age of AI search" (AI Discovery Survey, August 2025, n=1,927). Published October 16, 2025.
9. ChannelEngine / PR Newswire, "Only 17% of Consumers Trust AI Enough to Complete a Purchase" (Marketplace Shopping Behavior Report 2026, 4,500 shoppers). Published January 7, 2026.
10. Product.ai, "The 2026 Trust in AI Commerce Report" (n=1,463 US online shoppers). Fielded April 2026.
11. 5W Public Relations / PR Newswire, "5W Releases AI Platform Citation Source Index 2026" (680M+ citations). Published May 1, 2026.
12. Profound, "AI Platform Citation Patterns" (historical Reddit and Wikipedia shares). Published August 2025.
13. 5W Public Relations, "Supplements AI Visibility Index 2026" (3,800 supplement prompts, Jan-Apr 2026; four-brand 47%+ concentration). Published 2026, refreshed quarterly.
14. Coherent Market Insights, "Infrared Saunas Market Size, 2026-2033." Published 2026.
15. Grand View Research (via Hello Red Light), "Red Light Therapy Market Size Report 2026-2033." Published 2026.
16. Leapd, "How ChatGPT, Google AI Overviews, and Perplexity Source Information in 2026." Published January 2026.
17. Memeburn, "Google AI Overview Statistics 2026." Published 2026.
18. Search Engine Land / TechCrunch (via Digital Elevator compilation), OpenAI 900M weekly active users. Published February 27, 2026.
19. OpenAI Help Center, "Shopping with ChatGPT Search." Accessed 2026.
20. Search Engine Journal, "Lessons Learned From Adobe's 2026 Q2 AI Traffic Report." Published 2026.
21. McKinsey & Company, "The trends defining the $1.8 trillion global wellness market" (82% of US consumers call wellness a top or important priority). Published November 2024.
22. Circana via Fitt Insider, "US consumers prioritizing well-being wield $1.1T in spending power." Published April 15, 2025.
23. Fitt Insider, "Health & Wellness 2026 Halftime Report." Published July 14, 2026.
24. Reuters, "Reddit sues Perplexity for scraping data to train AI system" (SDNY filing). Published October 22, 2025.
25. 5W Public Relations, "The State of AI Citations 2026" (680M+ citations; Wikipedia ~47.9% of ChatGPT's top-10 source share). Published June 2, 2026.
26. Discovered Labs, "AI Citation Patterns: How ChatGPT, Claude, and Perplexity Choose Sources" (citation volume by platform; ~12% overlap with Google's top 10). Published December 2025.
27. Tinuiti via MediaPost, "Google AI Cites More Social, As Perplexity Pulls Back From Reddit" (Reddit falls from 25% of Perplexity citations in February to 7% in April 2026). Published May 27, 2026.

---

Note on source quality: Vendor-produced indices (5W Public Relations, Profound, Leapd, Discovered Labs) are marketing research from firms that sell AI visibility services, and their figures are self-described as observed, estimated, or modeled rather than peer-reviewed. They are cited here as directional evidence and attributed by name. Primary research from Adobe Analytics, McKinsey, Euromonitor International, Muck Rack, Product.ai, Semrush, Circana, and the Global Wellness Institute anchors the core claims. Platform citation shares are volatile and measured in weeks; the appendix date-stamps each figure for that reason.