UK search
& AI visibility
Where customer search interest is heading, who gets found on Google, who gets recommended by ChatGPT and Google AI Mode, and what specialist jewellers should do next.
Editor's note
Welcome to the first edition of our Jewellery State of Search report. This edition looks at H1 2026 vs H1 2025. Future editions will be run quarterly.
Google rankings and AI answers have been captured in September. Across the 317 buying searches we track, interest was flat compared with the same period last year. But that headline hides some substantial differences between individual brands and product categories: specialist jewellers grew while the chains and luxury houses declined.
We wanted to understand those differences: which brands are gaining ground, where customer demand is changing, and what other jewellery retailers can learn from them.
AI is also coming up more often in our conversations with brands. Alongside the search analysis, we’ve taken a separate snapshot of AI visibility to explore which businesses appear in AI-generated answers. One finding stands out: the leaders in organic search are not always the leaders in AI visibility.
For ecommerce teams, this raises some useful questions. Are you reaching customers who don’t already know your name? Are you visible in the categories where demand is growing? And which competitors deserve a closer look?
We created this report to help you answer those questions and identify where to focus your effort next. I hope you find it useful.
Five of the biggest findings from this edition
- +22%Specialists are taking the sector from the chains and luxury housesSpecialist jewellers' UK organic traffic rose 22% vs the first half of 2025, while chains were flat and luxury houses fell 24%. Specialists also hold 46% of Google's top-ten results across the 317 searches and take 77% of ChatGPT's recommendations.
- ±2%Search interest for fine and diamond jewellery was flat in the first half of 2026The 317 searches we track added up to about 437,000 searches a month, almost exactly the same as a year earlier. Despite that, there have been some interesting changes in specific keywords and keyword clusters.
- 57%A specialist holds the widest organic coverage, but five of the ten most visible domains are chainsDiamonds Factory reaches the top ten on 57% of searches. Beaverbrooks covers 30% of searches but 52% of search demand, because it ranks in the top ten for the biggest head terms, including engagement rings itself.
- 30%ChatGPT's most recommended jeweller ranks in Google's top ten for just 2% of searchesFlawless Fine Jewellery is put forward in 30% of ChatGPT answers tracked. Because ChatGPT names about nine businesses per answer, that is a 3.4% share of all its recommendations. Google AI Mode instead names four of the organic top eight.
- 4%AI buying answers cite retailers' own pages, not publicationsPublishers, forums and review sites supply about 4% of citations on ChatGPT and 3% on AI Mode. Reddit is the only editorial source in AI Mode's top ten.
How is customer search interest changing?
Decision it supports: which categories deserve closer attention in merchandising, content and campaign planning?
| Product category | Monthly search volume | Change |
|---|---|---|
| Engagement rings | 193,277 | 0% |
| Coloured gems | 87,584 | −5% |
| Bracelets | 64,758 | +4% |
| Earrings | 50,241 | −5% |
| Necklaces | 33,024 | +6% |
| Bespoke | 7,941 | −2% |
| lab diamond tennis bracelet | Bracelets | Lab-grown diamond | 928 | +100% |
| diamond pendant necklace | Necklaces | Natural diamond | 1,871 | +87% |
| lab diamond necklace | Necklaces | Lab-grown diamond | 569 | +86% |
| diamond stud earrings for women | Earrings | Natural diamond | 1,042 | +66% |
| platinum diamond engagement ring | Engagement rings | Natural diamond | 724 | +58% |
| diamond earrings women | Earrings | Natural diamond | 654 | +54% |
| lab grown diamond necklace | Necklaces | Lab-grown diamond | 814 | +43% |
| natural diamond engagement rings | Engagement rings | Natural diamond | 906 | +42% |
| yellow diamond engagement ring | Engagement rings | Natural diamond | 855 | +38% |
| custom engagement ring | Bespoke | Natural diamond | 580 | +30% |
| diamond cross necklace | Necklaces | Natural diamond | 882 | −31% |
| lab grown engagement rings | Engagement rings | Lab-grown diamond | 1,924 | −31% |
| gold diamond necklace | Necklaces | Natural diamond | 620 | −32% |
| diamond necklace uk | Necklaces | Natural diamond | 708 | −33% |
| sapphire and diamond engagement ring | Coloured gems | Sapphire | 680 | −36% |
| design your own engagement ring | Bespoke | Natural diamond | 693 | −36% |
| ruby and diamond ring | Coloured gems | Ruby | 1,036 | −36% |
| three diamond engagement ring | Engagement rings | Natural diamond | 502 | −37% |
| 3 diamond engagement ring | Engagement rings | Natural diamond | 633 | −43% |
| morganite engagement ring | Coloured gems | Morganite | 525 | −50% |
| Tennis bracelet | Bracelets | 2 | 35,626 | +9% |
| Diamond stud earrings | Earrings | 5 | 7,338 | +7% |
| Oval engagement rings | Engagement rings | 2 | 11,056 | +5% |
| Gold engagement rings | Engagement rings | 5 | 13,123 | +4% |
| Diamond necklace | Necklaces | 3 | 14,418 | +4% |
| Engagement rings | Engagement rings | 7 | 107,780 | +2% |
| Diamond bracelet | Bracelets | 4 | 8,966 | 0% |
| Emerald ring | Coloured gems | 3 | 11,089 | −1% |
| Diamond engagement rings | Engagement rings | 14 | 15,603 | −3% |
| Emerald engagement rings | Coloured gems | 2 | 8,245 | −4% |
| Sapphire ring | Coloured gems | 5 | 9,588 | −7% |
| Sapphire engagement rings | Coloured gems | 3 | 6,152 | −7% |
| Diamond earrings | Earrings | 10 | 21,873 | −9% |
Figures are Ahrefs estimates of monthly searches, January to June 2026 compared with the same months of 2025. They measure interest, not sales or customers. Each search sits in one product category. Diamond searches are counted as natural unless they say lab-grown, which is how the trade reads them, because a shopper who wants a natural diamond rarely types the word. Some of those shoppers will end up buying lab-grown, so the natural figures are a ceiling. A search that does not mention a budget or style is never treated as a luxury buyer. Ahrefs changed how it estimates volumes in September 2025, and every change on this page is adjusted for that, as the methodology explains. After the adjustment, engagement rings and necklaces could be read as either slightly up or slightly down, so treat both as flat. Every other category moves the same way whichever way the adjustment is made.
A modern twist on a traditional industry
An interesting change is taking place amongst some of the bigger keywords. "Tennis bracelets" (~34k searches per month) grew by 9% year-on-year, which is probably no surprise to anybody in the industry.
"Diamond earrings" on the other hand declined by 7% YoY which was surprising to see as it's usually a mainstay of the industry.
We're seeing search terms shifting to more and more specific terms, for example "diamond pendant necklace" growing search volume by 87%.
The lab grown segment of the market is continuing to go from strength to strength, with a lot of the keywords posting double-digit growth.
Who are the main players in organic discovery?
Decision it supports: are we competing with who we think we're competing with, or are other businesses actually the ones getting found first?
| Business type | Sites counted | Organic share of voice | UK organic traffic change | Brand-name search change | Share of ChatGPT recommendations | Share of AI Mode recommendations |
|---|---|---|---|---|---|---|
| Specialist | 26 | 46% | +22% | +4% | 77% | 52% |
| Chain | 17 | 34% | 0% | −8% | 16% | 36% |
| Luxury house | 4 | 10% | −24% | −4% | 6% | 12% |
| diamondsfactory.co.uk | Specialist | 57% | 14% | 61% | −5% |
| tiffany.co.uk | Luxury house | 44% | 22% | 34% | −34% |
| ernestjones.co.uk | Chain | 37% | 13% | 24% | −38% |
| thediamondstore.co.uk | Specialist | 33% | 14% | 25% | +95% |
| beaverbrooks.co.uk | Chain | 30% | 8% | 52% | +1% |
| austenblake.com | Specialist | 28% | 15% | 46% | +462% |
| diamond-heaven.co.uk | Specialist | 25% | 8% | 15% | +19% |
| goldsmiths.co.uk | Chain | 22% | 6% | 36% | −11% |
| laingsuk.com | Chain | 20% | 4% | 32% | +19% |
| rox.co.uk | Chain | 20% | 2% | 18% | +1% |
| Domain | Top 10 coverage |
|---|---|
| diamondsfactory.co.uk | 57% |
| tiffany.co.uk | 44% |
| thediamondstore.co.uk | 33% |
| austenblake.com | 28% |
| diamond-heaven.co.uk | 25% |
| debeers.co.uk | 15% |
| 77diamonds.com | 10% |
| caratlondon.com | 10% |
| novitadiamonds.co.uk | 10% |
| hkjewellery.co.uk | 8% |
| The Diamond Ring Company | Specialist | 300 | +58% |
| All Diamond | Specialist | 180 | +44% |
| Novita Diamonds | Specialist | 1,357 | +39% |
| Berry's Jewellers | Chain | 1,238 | +30% |
| Regal Hatton Garden | Specialist | 495 | +28% |
| Harriet Kelsall | Specialist | 1,225 | +19% |
| Queensmith | Specialist | 2,189 | +15% |
| The Diamond Store | Specialist | 16,436 | +10% |
| Beaverbrooks | Chain | 74,041 | +5% |
| ROX | Chain | 3,618 | +4% |
| De Beers | Luxury house | 2,884 | +1% |
| Diamond Heaven | Specialist | 3,823 | +1% |
| Ernest Jones | Chain | 19,563 | −41% |
| Fraser Hart | Chain | 1,648 | −13% |
| Goldsmiths | Chain | 13,720 | −11% |
| Austen Blake | Specialist | 1,377 | −11% |
| Warren James | Chain | 138,956 | −9% |
| Lily Arkwright | Specialist | 1,676 | −7% |
| Pravins | Chain | 262 | −7% |
| 77 Diamonds | Specialist | 3,871 | −6% |
| Diamonds Factory | Specialist | 2,674 | −5% |
| Tiffany | Luxury house | 59,263 | −4% |
| Laings | Chain | 2,863 | −3% |
| Chisholm Hunter | Chain | 10,094 | −2% |
| Glamira | Specialist | 3,827 | −2% |
| Mappin & Webb | Chain | 3,041 | −1% |
Google's standard organic results only, on the same 317 searches for every domain, as ranked on 10 September 2026. AI features are counted separately below. Filtering the peer view does not change a score: 57% means 57% of all 317 searches. Search demand covered weights each search by how often it is searched, so it favours domains that rank for the big head terms. It is a supporting measure, not market share. UK organic traffic change is Ahrefs' estimate for the whole website, not just the jewellery searches in this report, so it moves for reasons outside this report. Domains under 1,000 estimated visits a month show n/a, because a percentage on a few hundred visits means little. Ahrefs changed how it estimates volumes in September 2025 and traffic estimates are built from those volumes, so part of each change may be that revision rather than real movement. Read small changes with care. Brand-name searches are people typing the brand's name into Google: a measure of demand for the brand, not of visits to its site. In the business type table, organic share of voice is each type's share of every top-ten result across the 317 searches, with the balance held by marketplaces, publishers and unclassified sites. Traffic covers the 60 largest domains in the field and brand-name searches the 30 brands covering the most demand, with marketplaces and general retailers left out of both. The AI columns are each type's share of the 600 recommendations on each platform.
The generic brands are suffering.
A working theory I have is that AI has trained us to be more specific in what we're looking for and this behavioural change is helping the specialist brands and luxury houses that have their strategy right, but not so much the more generic, mass-market brands such as Ernest Jones, Goldsmiths and Warren James.
The lesson here: satisfy the more specific search intent rather than just focusing on the category leading search terms.
Which categories and pages are competitors winning?
Decision it supports: where is our most relevant visibility gap, and what kind of page should we investigate?
| Domain | Engagement rings | Earrings | Necklaces | Bracelets | Bespoke | Coloured gems |
|---|---|---|---|---|---|---|
| tiffany.co.uk | 26% | 23% | 31% | 31% | 0% | 12% |
| austenblake.com | 21% | 22% | 16% | 0% | 15% | 12% |
| diamondsfactory.co.uk | 16% | 20% | 7% | 22% | 0% | 12% |
| thediamondstore.co.uk | 9% | 12% | 18% | 24% | 0% | 15% |
| ernestjones.co.uk | 11% | 15% | 9% | 9% | 11% | 20% |
| beaverbrooks.co.uk | 7% | 7% | 11% | 18% | 0% | 5% |
| diamond-heaven.co.uk | 4% | 17% | 7% | 0% | 26% | 2% |
| goldsmiths.co.uk | 3% | 7% | 18% | 9% | 0% | 0% |
These pages show what ranks. Nothing here claims a particular design, content feature or technical change caused the result.
Which jewellers make the AI shortlist?
Decision it supports: does our visibility in Google translate into being recommended by AI, and which competitors appear instead?
| Brand | Share |
|---|---|
| Flawless Fine Jewellery | 3.4% |
| Diamond Heaven | 3.4% |
| The Diamond Ring Company | 3.1% |
| Hatton Garden Diamond | 2.9% |
| Goldsmiths | 2.6% |
| Queensmith | 2.4% |
| All other brands (523) | 82.2% |
| Brand | Share |
|---|---|
| The Diamond Store | 7.3% |
| Goldsmiths | 6.9% |
| Beaverbrooks | 6.4% |
| Ernest Jones | 6.1% |
| Queensmith | 4.2% |
| Mappin & Webb | 3.8% |
| All other brands (233) | 65.3% |
| Flawless Fine Jewellery | Specialist | 2% | 30% | 3% | 1% | 2% |
| Diamond Heaven | Specialist | 25% | 29% | 6% | 6% | 6% |
| The Diamond Ring Company | Specialist | 2% | 27% | 1% | 1% | 1% |
| Hatton Garden Diamond | Specialist | 0% | 25% | 4% | 7% | 2% |
| Goldsmiths | Chain | 22% | 22% | 28% | 17% | 16% |
| Queensmith | Specialist | 5% | 21% | 17% | 2% | 7% |
| Austen Blake | Specialist | 28% | 20% | 6% | 2% | 4% |
| F.Hinds | Chain | 8% | 18% | 7% | 12% | 3% |
| Diamonds Factory | Specialist | 57% | 17% | 7% | 4% | 8% |
| Purely Diamonds | Specialist | 4% | 16% | 1% | 3% | 0% |
| Ernest Jones | Chain | 37% | 16% | 25% | 10% | 13% |
| The Diamond Store | Specialist | 33% | 14% | 30% | 7% | 23% |
| 77 Diamonds | Specialist | 10% | 12% | 9% | 2% | 4% |
| H.Samuel | Chain | 16% | 10% | 8% | 3% | 3% |
| Mappin & Webb | Chain | 14% | 10% | 16% | 1% | 6% |
| Boodles | Luxury house | 0% | 8% | 9% | 4% | 2% |
| Taylor & Hart | Specialist | 1% | 8% | 10% | 4% | 4% |
| Beaverbrooks | Chain | 30% | 7% | 26% | 2% | 16% |
A recommendation is a business the answer puts forward as somewhere to buy. A mention names it without recommending it. A citation is a link the answer showed. The three are counted separately, and the donuts show recommendations only. In the table, each percentage is the share of that platform's 600 answers in which the brand was recommended, or in which its website was cited. ChatGPT names about nine businesses per answer and spread its recommendations across 529 different businesses, with the top six taking 18% between them. AI Mode names about four per answer, spread across 239, with the top six taking 35%. Each question was asked three times to test consistency, not to stand for three shoppers: the most recommended AI Mode brand was named in all three runs on only 32 of the 200 questions, ChatGPT's on 25. A 200-question panel is too small for statistical claims about a single product category. Collected through DataForSEO, set to the UK, in English, on 10 September 2026.
A strong hint that GEO isn't the same as SEO
A brand leads the way in ChatGPT share of voice that only has a top 10 ranking for 2% of the 317 keywords we track in traditional search.
As you'd expect, we've had to adapt our own SEO offering to include AI discoverability (called GEO/AI SEO by some). In the research we've done, AI discoverability shares a lot of things with traditional SEO, but there are also some big differences.
One of the biggest differences we've observed is that you need to be mentioned by third-party publications and you need to be creating well thought out, deep expertise pieces around the jewellery you sell.
Which sources appear in AI buying answers?
Decision it supports: which editorial or partnership opportunities deserve investigation, and which customer questions lack useful coverage?
| goldsmiths.co.uk | ChatGPT | Chain | 147 | 63 | Coloured gems, Necklaces | bespoke-lab-grown-diamond-jewellery |
| templeandgrace.co.uk | ChatGPT | Specialist | 142 | 51 | Coloured gems, Bracelets | jewellery/diamond-bracelets |
| hattongardenbespokejewellery.com | ChatGPT | Specialist | 113 | 43 | Bespoke, Coloured gems | pages/rings |
| ernestjones.co.uk | ChatGPT | Chain | 100 | 39 | Earrings, Engagement rings | bloom-rome-14ct-white-gold-550ct-lab-gr… |
| fhinds.co.uk | ChatGPT | Chain | 98 | 39 | Coloured gems, Necklaces | design-service/Bespoke-Diamond-Eternity… |
| regalhattongarden.co.uk | ChatGPT | Specialist | 77 | 38 | Engagement rings, Bracelets | collections/natural-diamond-engagement… |
| thediamondstore.co.uk | AI Mode | Specialist | 221 | 53 | Necklaces, Bracelets | collections/tennis-bracelets |
| beaverbrooks.co.uk | AI Mode | Chain | 183 | 42 | Necklaces, Bracelets | jewellery/bracelets/tennis-bracelets |
| ernestjones.co.uk | AI Mode | Chain | 114 | 30 | Bracelets, Necklaces | l/white-gold-diamond-bracelets |
| queensmith.co.uk | AI Mode | Specialist | 100 | 24 | Engagement rings, Bespoke | homepage |
| diamondsfactory.co.uk | AI Mode | Specialist | 91 | 31 | Bracelets, Earrings | homepage |
| goldsmiths.co.uk | AI Mode | Chain | 89 | 27 | Necklaces, Earrings | c/Jewellery/Gemstones/Diamond/Diamond-B… |
Every ChatGPT question and 183 of the 200 AI Mode questions drew several sources across their three runs. Diamond earring, necklace and bracelet questions drew the most consistent retailer citations on both platforms.
Bespoke questions are thinly sourced: 8 of the 32 drew one source or none across their three AI Mode runs, against 3 of the 40 engagement ring questions. Editorial coverage is thin everywhere. Publishers, forums and review sites supply about 4% of citations.
A page the AI read while answering is not counted unless the answer showed it as a source. Being cited often does not prove that coverage on that site causes a recommendation. AI answers change between runs and between weeks.
Citations are the links each answer actually showed: ChatGPT's source list and AI Mode's reference panel. A page ChatGPT read while searching is not counted unless it appeared in the answer. Being cited often does not prove that coverage on a site causes a recommendation.
What should a specialist jeweller prioritise?
Decision it supports: what should we ask our team or agency to focus on next? Sector-wide, not a brand audit.
Track ChatGPT and Google AI Mode separately, because they shortlist differently
- Evidence
- AI Mode names established organic performers: The Diamond Store appears in 30% of its answers and is cited in 23%. ChatGPT sends 77% of its recommendations to specialists, led by Hatton Garden independents such as Flawless Fine Jewellery and Hatton Garden Diamond, which rank in Google's top ten for 2% of searches or none at all.
- Next investigation
- Run the panel questions for your categories and record where your brand is recommended, only mentioned, or cited, platform by platform. Compare that with your organic coverage on the same searches.
- Limitation
- Even the leading brand was recommended in all three runs on only 32 of 200 questions, so a single answer proves little. Answers also drift from week to week.
Compete where the field is thin: bespoke and coloured gemstone
- Evidence
- Only two domains hold top-three coverage of 15% or more in coloured gemstone, and the bespoke leaders are small independents rather than the chains and diamond specialists that dominate every other category. Bespoke questions were also the thinnest in AI Mode citations, with 8 of 32 questions drawing one source or none across three runs.
- Next investigation
- Map your bespoke and gemstone pages against the searches in those two categories and list the searches with no relevant landing page. Check whether your bespoke process page is being cited at all.
- Limitation
- Bespoke interest is small, about 7,900 searches a month across all 317 searches. Coloured gemstone is large but fell 5% on the first half of 2025, so the opportunity is taking share, not riding growth.
Be clear whether you sell lab-grown, natural or both, because AI shortlists split on it
- Evidence
- When a question asks for lab-grown, AI Mode names The Diamond Store, Queensmith and Taylor & Hart. When it asks for natural, it names the chains. Lab-grown is still the small segment: about 6,700 engagement ring searches a month say lab-grown, against about 187,000 that do not say either way. Lab-grown engagement ring searches fell about 18% on the first half of 2025, while lab-grown necklace, bracelet and earring searches rose.
- Next investigation
- Check that each collection page says clearly whether its stones are natural or lab-grown, and test whether your brand is named on the lab-grown and natural questions in the panel. Decide which of the two you want to be recommended for.
- Limitation
- Search interest is not sales. Most diamond searches do not say natural or lab-grown, so the natural figure is a ceiling, and the lab-grown changes rest on small search volumes.
Methodology and evidence
- Geography
- UK search data. AI answers collected with a UK location, in English, on desktop.
- Organic sample
- 317 non-branded buying searches across six product categories, fixed on 10 September 2026. They were gathered from Ahrefs' matching terms for 36 seed phrases and kept where they have at least 100 searches a month. Interest is the average monthly volume for January to June 2026, compared with January to June 2025. Excluded: brand names and collections, resale, simulants, silver, wedding bands and informational phrasing. Brand websites stay in the field wherever they rank for generic searches.
- Competitive field
- Every domain with a top-ten Google result on any of the 317 searches, as ranked on 10 September 2026. No domain is excluded for being small. The peer view shows the specialists and luxury houses from that same field, scored the same way.
- AI panel
- 200 shopping questions that ask which retailer to buy from, 30 to 40 per product category, with no brand names and locations limited to the UK and London (panel version 1). Each was asked three times on ChatGPT and three times on Google AI Mode through DataForSEO on 10 September 2026: 1,200 answers in all. AI answers are a snapshot on the day and cannot be backdated.
- AI Overviews
- Four of the 317 searches returned a Google AI Overview on 10 September 2026. AI Overviews are counted separately from the organic results and from the AI panel.
- Counting rules
- A recommendation is a business the answer puts forward as somewhere to buy. A mention names it without putting it forward. A citation is a link the answer showed. The three are never added together, and Google and AI figures are never combined.
- Sources and tools
- Ahrefs Keywords Explorer, SERP Overview and monthly volume history. DataForSEO ChatGPT LLM Scraper and Google AI Mode SERP API.
- Branded search
- Monthly UK searches for each brand name, January to June 2026 against the same months of 2025, for the thirty brands covering the most search demand across the 317 searches. Marketplaces and general retailers are excluded, because searches for their names are mostly not about jewellery. Brands under 100 searches a month are left out, since a percentage on that base means little.
- Volume revision
- Ahrefs changed how it estimates search volumes in September 2025. Across these 317 searches, those below about 2,500 a month jumped by a median of about 1.25 times at that point, while larger searches did not move. Left uncorrected, a comparison of 2026 with 2025 would read that jump as growth. Each search is corrected by its own measured jump, taken from how it moved from August into September and October 2025 compared with the same months of 2024, which cancels the usual autumn rise. Brand-name searches showed no jump and are not corrected. Individual corrections are noisy, so changes for searches under 2,000 a month are a direction rather than an exact figure. Related searches are also grouped into product clusters, shown at 5,000 searches a month or more.
- Unavailable comparisons
- This is the first edition, so there is no earlier coverage to compare against: these figures are the baseline. In its place the field table carries each site's whole-site UK organic traffic change, which moves for reasons beyond the searches in this report.
- Refresh
- Quarterly, on the same searches, panel and counting rules.
Would you like us to look into your brand?
If you're looking at moving agencies, we'd be more than happy to look at your brand across SEO, AI SEO and Paid Search.
Book in for a discovery call today.