Rings.com Research · Edition 5
AI gave you a list of jewelry brands. Here's what that list actually is.
September 30, 2026 · ChatGPT, Gemini, Perplexity, Claude + Google AI Overviews · 30 queries · 3,000 responses
You asked an AI where to buy an engagement ring. You got a confident, well-formatted answer. Here is what you did not get: a neutral one. Five months into this study, the lists can still move a lot in a single month: one comparison site, Rare Carat, roughly doubled its AI visibility score (how often and how early the AIs list a brand; its raw mentions rose 63%), Brilliant Earth held the top spot, and James Allen, which now redirects to Blue Nile, still appears in more than half of the answers without most of those answers saying so.
Rings.com is an independent research platform and also a participant in the market it measures: we operate Ring Advisor, our own free ring-shopping guidance tool built in part on this study's data. So we hold ourselves to this study's own test. Rings.com is tracked under the same methodology as every brand here, and through all five editions it has recorded zero mentions. If that ever changes, we will say so in that edition. We have no financial relationship with any brand in this study. This is Edition 5; we run the study every month.
This page summarizes the key consumer findings. Full methodology and brand rankings are in the complete research report.
Rare Carat, a diamond shopping site, roughly doubled its AI visibility score (how often and how early the AIs list a brand) in one month: 3,494 points in August, 7,051 in September (+102%). The score is not a mention count: raw mentions rose 63% (530 to 862). It is the largest single-month move in points and in standard deviations this study has measured, and it is beyond measured noise (in percent terms a smaller brand, Keyzar, rose more). About 62% of the added mentions are in Perplexity, whose answers also became much more uniform from run to run, so the noise test, which measures scatter within a single session, understates how much that surface varies; leaving Perplexity out, the score rose about 69%. ChatGPT named Rare Carat in just 10 of its 600 responses. This is a one-month observation, and the study can describe the change but not explain it.
Two of the four mall chains, Jared (down 44.9%) and Helzberg (down 55.5%), are still below their May AI visibility scores by more than measured noise. Kay Jewelers rose this month and is now within noise of its May level, and Zales fell back after August's rebound and is also within noise of May.
Heritage luxury brands like Cartier, Harry Winston, and Tiffany made up 15.3% of ChatGPT's brand mentions this month versus 5.8% of Gemini's, the highest and lowest of the four models, a gap of about 2.6 times across the same queries. In August the gap was 3.0 times.
Google's AI Overview, the AI answer box at the top of a Google search, appeared on 564 of the 600 search checks we ran this month (94.0%). In August it was 520 of 600 (86.7%).
The mall-chain percentages compare each brand's September weighted score with its May baseline, with the 20-run month divided by two so it matches May's run count; each comparison is tested against the brand's own measured run-to-run noise (Jared and Helzberg are beyond the strict 3.5 SD bar, Kay Jewelers and Zales are within noise). The Rare Carat figures are raw weighted scores from the two 20-run months. The luxury figure is that category's share of each model's total brand mentions this edition. Heritage luxury: Cartier, Harry Winston, Tiffany & Co., David Yurman. Mall chains: Zales, Jared, Kay Jewelers, Helzberg. One update from August: the models' disagreement about mall chains, 2.5x in August, was 1.8x this month, and which models sit highest and lowest changed. The full by-model breakdown is in the research report.
Jump to what you should do differently →
We ran 30 shopping questions through ChatGPT, Gemini, Perplexity, and Claude, plus Google's AI Overviews (the answer box at the top of a Google search), across 3,000 responses this month. The same question asked to different AI tools still returned structurally different brand lists, in ways that would send you to entirely different types of stores, at entirely different price points, before you have looked at a single ring. And this month Brilliant Earth kept the #1 spot over Blue Nile by a margin our noise test says is beyond random noise, as in August, while Rare Carat climbed from 8th to 4th by score.
The AI tool you happen to use first shapes your entire consideration set. That choice is invisible, and the set it gives you is quietly changing month to month.
What that looks like in practice
Four of the 30 queries from this month's study, scored across twenty runs per model. Each card shows the six brands named in the most of the 20 runs (and named in at least 2); brands further down are not shown. Ties are broken by the brand's position-weighted score on that question, and any still tied by which appeared first in the data, so an equally named brand can fall just outside the six. (A brand listed 11th or later in an answer still earns 1 point.)
ChatGPT
Cartier
Harry Winston
Blue Nile
James Allen
De Beers
Brilliant Earth
Gemini
Brilliant Earth
Blue Nile
James Allen
Frank Darling
Harry Winston
Cartier
Perplexity
Blue Nile
Whiteflash
Brilliant Earth
Rare Carat
Cartier
VRAI
Claude
Tiffany & Co.
Harry Winston
Blue Nile
Brilliant Earth
Cartier
James Allen
The same question, four different reads. Claude names Tiffany first in all twenty runs, so its top "alternative to Tiffany" is Tiffany itself. Gemini also names Tiffany first in 14 of 20 runs, but it names Tiffany in fewer runs than six other brands, so Tiffany is not among the six shown. ChatGPT puts Cartier first in 16 of 20 runs and lists James Allen, a brand that now redirects to Blue Nile, in all twenty. Perplexity has no single leader (Blue Nile first in 6 runs, Whiteflash in 5) and named Rare Carat in seventeen of twenty runs; in August no model named Rare Carat for this question at all.
ChatGPT
James Allen
Blue Nile
Brilliant Earth
Ritani
Whiteflash
Tiffany & Co.
Gemini
Blue Nile
Brilliant Earth
Clean Origin
Whiteflash
Rare Carat
James Allen
Perplexity
Blue Nile
Rare Carat
Brilliant Earth
James Allen
Claude
Rare Carat
Blue Nile
Whiteflash
Brilliant Earth
Tiffany & Co.
Ritani
Rare Carat, a diamond shopping site, is the clearest example this month. It was named in 56 of the 80 responses to this question (27 in August). Perplexity and Claude named it in all twenty runs and Gemini in fifteen, while ChatGPT named it in one of twenty runs (10 of its 600 responses across all 30 questions this month, after zero in August). If you want to see Rare Carat, which app you ask changes whether it shows up at all. This study measures what the models say, not why, and none of this is a recommendation of any retailer. About 62% of Rare Carat's added mentions came from Perplexity (where its answers also became more uniform from run to run: named in all 20 runs for 13 questions, versus 2 in August), and several of this month's other large shifts are also concentrated in Perplexity, though not all (Kay Jewelers' rise is mostly in Claude, 6 to 62 mentions); the study cannot say why. On this question ChatGPT also still lists James Allen first in 12 of 20 runs.
Show two more examples
ChatGPT
Brilliant Earth
VRAI
Mejuri
Taylor & Hart
Blue Nile
James Allen
Gemini
Aurate
Bario Neal
Do Amore
Brilliant Earth
Clean Origin
VRAI
Perplexity
Bario Neal
Brilliant Earth
Taylor & Hart
VRAI
Do Amore
Mejuri
Claude
Brilliant Earth
VRAI
Bario Neal
Do Amore
Mejuri
Aurate
Adding one word, "ethical," pulls in a noticeably different brand set. Bario Neal, a jeweler that appears mostly on ethical-sourcing questions, is named in nineteen or twenty of twenty runs by Gemini, Perplexity, and Claude, but only four of twenty by ChatGPT, and it barely registers on general queries (216 of its 224 mentions come on the ethical-sourcing questions). The most relevant brands for what you actually care about may only appear when you ask the right question.
ChatGPT
Blue Nile
James Allen
Brilliant Earth
Costco
Whiteflash
Gemini
Blue Nile
Ritani
Zales
Rare Carat
Clean Origin
Jared
Perplexity
Zales
Angara
James Allen
Blue Nile
Rare Carat
Claude
Rare Carat
Blue Nile
Brilliant Earth
Ritani
Zales
Kay Jewelers
The mall names show up on the budget question. Perplexity names Zales in all twenty runs, Claude in nineteen, and Gemini in seventeen; Claude also names Kay Jewelers in sixteen. Zales' overall AI visibility fell back this month after August's rebound and is now within noise of its May level. ChatGPT names Costco in ten of twenty runs here; this study does not assess its prices.
Check before you click: James Allen still ranks third in this study, appearing in 1,295 of this month's 2,400 chat responses (54.0%), about the same as August's 1,300 (54.2%). It is not an independent store anymore. The brand has been merged into Blue Nile, and jamesallen.com now redirects to bluenile.com, yet most of the answers that include it show no acknowledgment of that which the keyword screen can detect (about nine in ten, 90.7%; the screen can miss some, see the report). If an AI recommends James Allen to you, know that you are being sent to Blue Nile.
The answer above your search results
Google AI Overviews, the AI-written answer box at the top of most Google searches, showed up in 564 of 600 search checks this month (94.0%) (30 questions, tested 20 times each; 520, or 86.7%, in August; 89.8% in July; and 283 of 300, or 94.3%, in June, which used 10 runs with 2 blank rows excluded). The share that came back with a list of sources was about the same as in August (518 of 600 versus 508); the difference is overviews that came back without one (46 versus 12).
And it is still a different list. Google's box again agrees with chat at the very top (Brilliant Earth, then Blue Nile) and ranks the diamond shopping site Rare Carat third, ahead of James Allen, the reverse of their order in chat (4th versus 3rd). Two brands rank in Google's top 10 while sitting outside the chat top 10 (both ranked by mentions): Bario Neal is 8th in Google's box versus 16th in chat, and Kay Jewelers 9th versus 14th.
Where Google's answer is really coming from
We could see the sources Google cited (this month a portion of the citation data arrived in an unreadable format, 727 of 4,468 entries, so these figures cover the citations we could resolve; details in the research report). The most-cited resolvable sources were
YouTube, Reddit, and Brilliant Earth's own site. Three of the top ten were community or video sites (YouTube, Reddit, and Quora). This is a one-month snapshot of what Google cited.
The practical takeaway: Google's jewelry answer cites community discussion alongside the brands' own pages. That can show what other shoppers say, but it is not a vetted ranking. Treat it the way you would treat a popular Reddit thread.
Three things to do differently
Ask two AIs, and glance at Google's box too
Paste the same question into ChatGPT and Gemini, and note what Google's AI Overview says. Three sources, a few minutes, meaningfully different brand sets. Names that appear in every source are the ones the AIs agree on, which is not the same as the ones that suit you. Google's box puts a couple of names, such as Bario Neal and Kay Jewelers, in its top 10 that the chat apps rank lower, so a cross-check can surface stores a single app would not show you.
Phrase toward your actual priority
"Best engagement ring brands" returns a generic list. "Best lab-grown diamond rings under $3,000" or "most ethical place to buy a diamond" return answers shaped by your real constraint. Vague questions return the same brands everyone gets. Specific ones surface different brands, such as Bario Neal on ethics queries.
Treat the list as names to research, not names to trust
There is no methodology behind AI recommendations. No consumer testing, no pricing analysis, no evaluation against your needs, and as the James Allen example shows, most of the answers that include it show no acknowledgment, that a keyword screen can detect, of the merger into Blue Nile. The lists also change from month to month: this month Rare Carat roughly doubled its score (while its raw mentions rose 63%), Kay Jewelers rose and Zales fell back. Use the list to build a research starting point, then check prices, reviews, return policies, and whether the company is still independently operating before deciding anything.
Canadian shoppers: you're getting a list of American stores
We track four major Canadian retailers (Birks, Michael Hill, Spence Diamonds, and People's Jewellers) in every edition. A single Michael Hill chat mention appeared in July; all four Canadian brands recorded zero in August and again this month, across 2,400 chat responses and Google's answer box. Practically speaking, the lists are American stores. This is a brand-tracking count, not a test of whether your location changes the answer.
Get future Rings.com Research in your inbox
We run this study every month. Get the next edition when it publishes. No other emails.
30 queries, 4 AI models plus Google AI Overviews, 20 collection runs, 3,000 responses, September 2026 (Edition 5). Full methodology in the complete research report. Raw data available on request. Rings.com receives no compensation from any brand in this study; it operates Ring Advisor, a ring-shopping guidance tool built in part on this study's data, and is tracked under the same methodology as every other brand (zero mentions in all five editions). Questions: rodney@rings.com