The Red Pin Geek AI Visibility Study: Does AI Recommend Independent Jewelers?
The one-line finding
Across three independent jewelry brands, two of mine and one a client who gave full permission, AI search engines cited the pages we built and stayed completely silent on the adjacent topics we had not built. A clean, repeatable split that says AI recommendations in this space are earned by specific content, not by brand size or luck.
Why I can say this when almost no one else can
Most "AI visibility" advice is opinion. This is a controlled measurement: on my own brands I split topics into treated (pages I actually built and invested in) and control (closely-matched adjacent topics I deliberately left untouched), then checked what AI engines cite. If the built topics get cited and the matched untouched ones do not, the content is doing the work, not the brand's size or history. That comparison is the whole point, and it is why the results are attributable rather than anecdotal. The client brand, Bohemi, comes in separately, as independent real-world confirmation through her own published results.
I run this on three independent jewelry brands. Two are mine, so I can publish every number: Red Pin Geek (my AI-visibility consultancy) and Andrea Li Designs (my one-of-a-kind jewelry brand). The third is a client, Bohemi, alternative engagement rings by Heather in Boulder, Colorado, included by name with her full permission. The client brand is the most important one here: it is external validation. A real, independent maker whose before-and-after results are already public in her own case study, so this is not me grading my own homework.
Method, in plain terms
- Engines checked: Perplexity and Gemini via a repeatable script, plus Google AI Overview and AI Mode, plus a manual ChatGPT glance. Perplexity is the reliable detector; a Gemini "no" is a floor, not a proven zero (it hides many citations behind redirect links a script cannot read). I say so rather than pretend the number is exact.
- Prompts: a fixed panel of buyer-style questions per brand (definition, comparison, "who makes / where to buy," brand-name), each run twice per engine so I can see run-to-run wobble.
- Treated vs control (my own brands): built topics vs matched adjacent topics never covered. On Andrea Li Designs the control is five gemstones I have no content for (topaz, turquoise, emerald, tanzanite, citrine); on Red Pin Geek it is adjacent SEO/AI territory I have not built (generic audits, Etsy, local service). Bohemi is not in this controlled panel; she comes in through her own published before-and-after results.
- Honest limits, stated up front: small first-party sample, a synthetic prompt panel, "how often" not "guaranteed," three brands not two hundred. It is a clean controlled read, not a census. That honesty is deliberate; a study that hides its limits invites the reader to distrust the parts that are solid.
Finding 1: Citations follow the content, not the brand
The clearest result is the control floor: across all three brands, every adjacent topic left unbuilt returned zero AI citations, while the topics that were built got cited repeatedly.
- Red Pin Geek (Day-0, 2026-07-17): cited in 5 of 7 built prompts (firmly, 2 of 2 runs in Perplexity on the core ones) and 0 of 5 control prompts. Clean floor.
- Andrea Li Designs (Day-0, 2026-07-17): the built guides, the pastel collection pillar, and the brand-name queries all got cited; the 5 control gemstones returned 0.
Same controlled pattern on both brands I own and measure directly: AI cites what you build and stays silent on what you don't. That is the quantified "Foundation Gap": most independent jewelers are absent from AI answers for their own categories not because AI dislikes small brands, but because there is no built page for the engine to cite. That absence is the default every new page starts from, and it is a content gap, not a verdict. The next finding shows what happens when the pages get built, including on a brand that is not mine.
| Brand | What we built (treated) | Control: matched topics left unbuilt | What AI did |
|---|---|---|---|
| Red Pin Geek (mine) | AI-visibility guides, the Foundation Gap pillar, tools | generic audits, Etsy, local-service SEO | Cited in 5 of 7 built prompts; 0 of 5 controls |
| Andrea Li Designs (mine) | gemstone guides, the pastel collection pillar | five gemstones with no content (topaz, turquoise, emerald, tanzanite, citrine) | Guides, pillar, and brand cited; 0 of 5 controls |
Bohemi, the client brand, is not in this controlled panel; her independent, published before-and-after appears in Finding 2.
Proof that "absent" is the starting default, not a slur: when I launched three new Andrea Li Designs category pages on 2026-07-19, two of them ("wearable art jewelry," "seashell jewelry") registered literally zero citations across all four engines on day one, the honest before-state every new page starts from.
The control floor
AI cited the pages I built, and nothing I left as a control
On two brands I own and measure directly, I checked AI answers for topics I had built pages for against closely matched topics I left untouched. The split was clean.
Every adjacent topic I never built a page for returned zero AI citations, on both brands. The content is doing the work, not the brand's size or history.
Source: Red Pin Geek and Andrea Li Designs first-party citation panels, Day-0 July 17, 2026 (Perplexity, the reliable detector). Bohemi's independent public results appear separately.
Finding 2: When a page gets cited, AI repeats your framing
Being cited is not just a link; the engine absorbs your explanation.
- On 2026-06-20, my Andrea Li Designs "aquamarine vs morganite" guide was cited in the Google AI Overview panel and ranked #1 organically, and the AI answer reproduced my own iron-vs-manganese explanation of why the two stones differ in color. The page did not just get found; it taught the machine the answer.
- On 2026-06-17, Google AI Mode cited Red Pin Geek three times in a single answer set.
- On Bohemi, the same mechanism shows up in an independent brand's published results. Heather built persona-and-style pages for exactly her niche, a "witchy engagement rings" collection, a boho collection, honest alternative-stone guides. Her case study shows the payoff: those searches did not rank for her at all before, and after she built the pages, "boho engagement rings" reached position 7 on page one, her witchy collection went from zero organic traffic to steady discovery, and site crawlability went from 68 of 650 pages to 740 pages. Bohemi is now picked up across Google's AI Overviews and AI Mode, with 92 of her pages cited by AI tools, all on organic content with no paid search. AI can recommend her for "witchy" and "celestial" rings because she built the only honest pages for them.
Independent validation · Bohemi, Boulder CO
She built the pages. Then search and AI found her.
Bohemi is a client, not one of my brands, so this is external proof. She built pages for her actual niche, and her published case study shows the before and after, all on organic content with no paid search.
Source: Bohemi's published case study, "How Bohemi Became an Engagement Ring Authority" (redpingeek.com), shared with the owner's permission. Public figures only.
Independent proof: a study found the same mechanism
If this were only my three brands, some skepticism would be fair. So here is independent proof of the mechanism. In 2024, a team of academic researchers (Aggarwal and colleagues, in the paper widely known as the Princeton GEO study) ran the first systematic test of whether content can be deliberately built to show up more often in AI answers. It can. The levers that worked best were adding real citations, statistics, and direct quotations to a page, which raised its visibility in AI-generated responses by up to 40 percent. They found in a formal study what my three brands show in the wild: AI does not reward the biggest name. It rewards the page that hands it something specific and trustworthy to quote.
A formal study found that adding real citations, statistics, and quotations to a page raised how often AI answers pulled from it by up to 40 percent. That is the same mechanism my three brands show in the wild.
Source: Aggarwal et al., "GEO: Generative Engine Optimization," 2024 (the Princeton GEO study), arXiv:2311.09735.
The takeaway for a jeweler: the page you write becomes the source the AI speaks from. Write the true, specific thing only you can say, and the engine repeats you.
Finding 3: Does adding authority to a commercial page move its citations? (Result publishing August 14, 2026)
Here is the cleanest test in the whole study, and I am publishing it before I know the answer, on purpose.
On my own brand, Andrea Li Designs, I took four commercial collection pages (pearl, aquamarine, tourmaline, and green amethyst) and gave each a deliberate boost in internal authority, meaning the links and signals that tell search engines a page matters. I left three closely matched collections (labradorite, moonstone, and garnet) completely untouched, as a control. At the start, all seven sat in the same spot: no clicks, buried in search, absent from AI answers.
The question is simple. Do the four I worked on start getting cited while the three I left alone stay flat? If they do, that is not a coincidence, it is cause, because the only thing that changed is the work. If they do not move, I will say that too.
I publish the result right here on August 14, 2026, win or lose. Subscribe below and I will send it to you the day it lands.
Finding 4: The pages that get cited are the ones that explain something
The pattern underneath all of this: educational guides and clear collection pages earned citations, while thin pages that mostly list products did not, even when they pulled real search traffic. On Andrea Li Designs, a couple of pages with big search impressions but little genuine explanation showed up constantly in Google yet stayed absent from AI answers, the "found but not chosen" gap. The lesson is not "publish more pages." It is "write the page that actually answers the question," because that is the one the AI can quote.
What gets cited
AI quotes the page that explains something
Guides and category pages that teachcited
Pages that answer a real question in language only the maker could write. These earned the AI citations.
Thin pages that just list productsnot cited
Even with plenty of search traffic, pages light on genuine explanation stayed absent from AI answers. Found, but not chosen.
Depth is the lever. The lesson is not "publish more pages," it is "write the page that actually answers the question," because that is the one AI can quote.
What this means for an independent jeweler
- If AI never mentions you, the likeliest reason is not that AI hates small brands. It is that there is no page of yours specific enough to cite. Absence is a content gap, not a verdict.
- Build the page only you can write, the real bench detail, the honest comparison, the process no competitor shares. That is what gets cited and repeated.
- Educational and category pages earn citations; thin product pages usually don't. Depth is the lever.
- You can check this on your own store the same way I did, starting with the free AI Visibility Score below.
Why it is worth the work: this is not a vanity metric. For an independent jeweler, being the name AI gives when someone asks for your kind of work is the difference between a ready-to-buy shopper finding you and never knowing you exist. For Bohemi, building this foundation tracked with a 53% rise in visitors from search and a 58% lift in sales, all on organic content. And unlike an ad, a citable page keeps earning: every page AI can quote is a salesperson that keeps introducing you long after you have hit publish.
Frequently asked questions
Does AI actually recommend small jewelry brands, or just the big names?
Yes, it recommends small brands, but only when there is a specific page for it to cite. In this study, three independent brands were each cited by AI for the exact topics they had built real pages for, and ignored on closely matched topics they had not. Size was not the deciding factor. A built, citable page was.
Why is my store invisible in ChatGPT and Google's AI answers?
Almost always because there is no page of yours specific enough for the AI to quote. AI answers are stitched together from pages the engine can find and trust on that exact question. If your site has products but no clear, honest page answering the question a shopper actually asks, the engine has nothing of yours to pull from, so it names someone else. That is a content gap, not a penalty.
What kind of page actually gets cited?
Depth beats thinness. In this data, educational guides and clear collection or "category" pages earned citations, while thin product pages with little real explanation did not, even when they had plenty of search traffic. The pages that win are the ones that answer a real question in language only the maker could write.
How do I check my own store?
Run the free AI Visibility Score. In plain language it shows you whether AI can currently find and describe your store, and where the gaps are.
Sources
- Andrea Li / Red Pin Geek first-party citation ledgers, 2026 (Red Pin Geek and Andrea Li Designs prompt panels, Day-0 2026-07-17).
- Bohemi (Heather, Boulder CO), included by name with the owner's permission: her published case study, How Bohemi Became an Engagement Ring Authority.
- Aggarwal, P., Murahari, V., Rajpurohit, T., Kalyan, A., Narasimhan, K., and Deshpande, A. (2024). GEO: Generative Engine Optimization. arXiv:2311.09735. The study finding that adding citations, statistics, and quotations raises a page's visibility in AI-generated answers by up to 40 percent.
Continue learning
- The Foundation Gap: Why AI Search Skips Independent Jewelers: the argument this study puts real numbers to.
- How Bohemi Became an Engagement Ring Authority: the full case study behind the Bohemi figures here.
- Score your own store on the free AI Visibility Score: the same "can AI find and describe you" check that started this study, run on your own store.
- Subscribe to the Red Pin Geek Substack (free): weekly, plain-language field notes on getting found by AI as an independent jeweler.
- Work with me on your own store (Snapshot / Full Audit / Full Audit + 1:1 call): I run this same built-vs-unbuilt read on your store and tell you exactly which pages AI can cite and which it cannot.
Andrea Li is the founder of Red Pin Geek, where she helps independent jewelry designers get found and recommended by AI search. She spent 18 years in the jewelry industry and runs her own one-of-a-kind jewelry brand, Andrea Li Designs. She writes about SEO and AI visibility at redpingeek.com and makes jewelry at andreali.com.
Want Red Pin Geek in Your AI Answers?
Google now lets you choose preferred sources that get labeled in AI Overviews and AI Mode. If this content helps you navigate AI visibility for your jewelry business, add us and you'll spot our links faster next time Google answers your questions.
Add Red Pin Geek as a Preferred SourceTakes one click. Works in Google Search, AI Overviews, and AI Mode.
