↓ Download PDF
Team Empathy  ·  Search & AI Strategy

Be the jeweller
AI recommends.

A Search and AI strategy for Ernesto Buono Fine Jewellery

Getting Ernesto Buono found on ChatGPT

Sydney · Australia   |   Prepared for William & Ernesto Buono  ·  August 2026

1. Executive summary

The short version. Organic search brings you about $263,000 a year through the appointment book, but $81,000 of that is branded, people searching your name. Only $182,000 is non-branded, from category searches, and that is the only part a search strategy can grow. Over the last 90 days you were shown in Google 612,410 times and earned 3,791 clicks, roughly one click for every 162 appearances, because you sit at position 26 rather than the top three. You are already visible across the whole Australian market. You are simply not being chosen.
Three things to know
The non-branded market, and the share of it you hold
4,729,200
non-branded searches a year across your categories in Australia
1,654,800
of those are organic clicks you could win, once ads and AI answers take their share
10,488
is what you win today, which is 0.6% of the non-branded market
$168,000
a year for every extra 1% you take, because 1% of the pool is 16,548 clicks

So the whole opportunity is one sentence: you hold 0.6% of the non-branded organic clicks in your category, and every extra 1% is worth about $168,000 a year. Your branded traffic is not part of this and does not change.

What that is worth

Built on your own numbers: an average order value of $7,000, a booking rate of 0.67% on organic traffic, and your July close rates of 37% in person in Sydney and a 15% goal over Zoom for the rest of Australia. Every column below is non-branded, and your $81,000 of branded organic revenue sits on top of all of it, unchanged. If the Zoom close rate lands above 15%, every figure here moves up with it. Modelled, plus or minus 40%

CaseShare of the non-branded marketNon-branded clicks / yrNon-branded revenue / yrNew revenue / yr
Today0.6%10,488$182,000—
Conservative1.5%24,822$327,000+$145,000
Mid3.0%49,644$579,000+$397,000
High5.0%82,740$915,000+$732,000
Where you focus, at a glance
Engagement rings
~51,600 searches / mo
Fix first
374 pages already live, 5 terms ranking, none in the top ten. Your core category and your emptiest result.
Lab grown
~24,900 searches / mo
Biggest gap
Only 37 pages against the fastest-growing demand in the category, sitting at an average position of 47.
National & remote
The Zoom play
Build new
Zero pages today. The interstate demand is already there and organised city by city.

Triangulated Australian search volume per territory, August 2026. The full keyword, prompt and cluster research behind every number here lives in The data tab. GKP + Ahrefs + SEMrush

2. How AI-ready you are today: the AEO scorecard

We score every brand out of 100 against our AEO Maturity Model: four pillars (Technical, Content, Authority, Measurement), each out of 25. Certified is 80 out of 100, Professional on all four, the standard we build clients to. You are at 45 today, with good bones and a clear climb.

Ernesto Buono45/100
Developing · good bones, no answer engine
Technical15/25
Content10/25
Authority15/25
Measurement5/25
+35
points to Certified. Technical and Authority are already at Structured. The whole climb sits in Content and Measurement, which are the two fastest pillars to move.
Where you are  ·  where you need to be
Ernesto Buono
45 / 100
Today · Developing, 45 / 100 Where you need to be · Certified, 80 / 100

The detailed findings, how the scoring works, and where every number comes from are in the appendix.

3. The market you are missing: search and AI

We sized the Australian market you sell into three ways, then narrowed it to what is realistically winnable.

StepVolumeWhat it means
The total category394,100 searches / moEvery non-branded term across engagement rings, wedding rings, lab grown diamonds, diamond education and fine jewellery in Australia. Celebrity searches such as "Taylor Swift engagement ring" are excluded, along with competitor brand names. Ahrefs
The measured core204,110 searches / mo112 terms triangulated across Google Keyword Planner, Ahrefs and SEMrush, taking the median of the three. This is 52% of the total and it is what the money model runs on, because every figure is verified by three independent tools. 3-tool median
The winnable click pool137,900 clicks / moRoughly 35% of searches survive to an organic click once ads, AI answers and zero-click results take their share. This is the pool everyone in your category competes for.
Your share of it today874 clicks / moYour non-branded organic clicks, which is 0.6% of the pool. Your brand searches already belong to you and we do not claim credit for them. GSC
The gap in one line. The category serves up 137,900 organic clicks a month in Australia and you take 874 of them. You already appear 204,000 times a month, so this is not about being invisible. It is about appearing at position 26 instead of position 3.
What that looks like on real terms

These are your own Search Console numbers from the last 90 days. Every one is a term you already show up for.

TermSearches
/mo
KDYour impressionsYour clicksYour position
lab grown engagement rings1,600135,055233.8
emerald cut diamond engagement rings59013,166022.9
emerald cut diamond ring2,90013,110016.4
engagement rings sydney2,400372,591517.3
lab grown diamond rings2,90092,552041.9
mens gold wedding rings45002,108030.9
lab grown diamonds11,00091,541063.8
eternity ring5,40001,498046.8
cushion cut engagement rings80001,491045.1
aquamarine engagement ring60001,351226.9

Search volumes are the median of Google Keyword Planner, Ahrefs and SEMrush. KD is Ahrefs keyword difficulty out of 100. Impressions, clicks and position are your own Search Console data for the 90 days to 1 August 2026.

Look at the difficulty column. Almost every one of these sits at a keyword difficulty of 0 to 13. These are not hard terms. "Emerald cut diamond ring" has 2,900 searches a month at a difficulty of 1, you appear 3,110 times, and you get nothing, because you sit at position 16. The one row where you are at position 6.9 is the one row earning clicks. That is the whole strategy in a single table: the terms are winnable, and being on page one is what changes the outcome.
The AI layer is live in your category now

Google AI answers already appear on 26 of the 112 terms we measured, including "lab grown diamonds", "eternity ring", "diamond clarity chart" and "how much to spend on an engagement ring". Your own GA4 shows an AI Assistant channel already sending Australian visitors, 63 sessions and 12 form submissions in 90 days, with nothing optimised for it.

We built you a library of 136 buyer questions across 13 territories, worth roughly 18,275 conversational searches a month. 127 of them are questions a bespoke Sydney atelier can answer better than a chain retailer can. That is the opening. Questions like "can I use my own diamond in a new setting", "how long does a custom ring take" and "can I design an engagement ring over a video call" have no good answer online in Australia today, and they are exactly what your buyers ask.

4. The opportunity, in dollars

The table in the executive summary is built from four numbers, all of them yours, and all of them non-branded. Here is the chain in full.

The searchesnon-branded, AU
4,729,200
The winnable clicksafter ads and AI answers
1,654,800
What you win today0.6% of the pool
10,488

Roughly one search in three survives ads, shopping carousels and AI answers as a winnable organic click, and you currently take one in every 158 of those. Modelled

Each new non-branded click is worth $10.14 to you. That is your 0.67% booking rate, multiplied by a close rate blended across Sydney and the rest of Australia, on a $7,000 order. That is the figure every case in the table multiplies.

A click you win today is worth more, about $17.36, because every appointment you take now is a Sydney one that closes in person at 37%. New demand skews interstate, where a Zoom consultation is assumed to close at 15%, so the blended value of new clicks is deliberately lower. We have modelled 30% of new demand as Sydney and 70% as the rest of Australia.

That is the whole model: win more of the 1,654,800 clicks.

The compounding curve (the mid case, arriving over twelve months)
$150k$200k$250k$300k$350k$400k$450k$500k$550k$600kYou own the systemTodayMonth 3Month 6Month 9Month 12$579K non-branded, the mid case

The mid case arriving, on non-branded revenue: certified by month 3, the system handed to you at month 6, all of it live by month 12. Hover any point for the maths. Modelled, directional

How the clicks get won

Two levers, both measured. Click-through on the impressions you already get, as you move from position 26 into the top three, which is where 612,410 impressions a quarter start converting instead of scrolling past. And more impressions, as the territories you have no page worth ranking for start to rank, chiefly lab grown and the interstate and remote-consultation questions. The conservative case assumes only the first lever really lands; the high case assumes both do.

Every figure here is non-branded and excludes repeat and referral business, which in a category built on referrals sits on top as a further slice. Your unchanged $81,000 of branded organic revenue sits on top of the non-branded column in the executive summary table.

One honest note on timing. Because a $7,000 engagement ring is researched for months rather than bought on impulse, the revenue in this model is genuinely back-loaded. Cumulative attributable revenue across the first six months lands near $33,000 on the mid case, while the annualised run-rate at month six is already up by about $139,000. Judge the first six months on run-rate and on the leading indicators, appointments booked and rankings gained, rather than on cash collected in period.

5. The six-month transformation

Three phases, in order. We move through them as the site is ready, not on a stopwatch.

PhaseWhat happens
Foundations
Make you readable
Fix the measurement first, because nothing else can be judged until it is right: filter the Meta data-centre traffic, rebuild the booking conversion so it fires once per real appointment, and repoint Google Ads at it. Unblock GPTBot and rewrite llms.txt properly. Attack the 14-second mobile LCP. Add FAQ schema and Review schema across the site. Clean up the duplicate H1s and the pages competing with each other.
Own your territories
Make you rank
Rebuild the category and collection pages that already draw impressions but sit at position 20 to 45, starting with engagement rings, lab grown and diamond shapes. Consolidate and upgrade the 127 articles you already have rather than adding to the pile. Build the answer layer for the 136 buyer questions, prioritising the 127 only you can answer well.
Authority and compounding
Make you the answer
Clean the link profile of the syndication and directory tail, then earn genuine Australian editorial coverage in the places that already cover you well. Build out the interstate and remote-consultation territory that has no pages at all today. Track AI visibility properly so you can see which assistants cite you and where.

The full scope, pricing, timeline and guarantee sit on The offer tab.

Appendix A: your channels and current position

Channels, Australia only, last 90 days
ChannelSessionsAppointmentsAppointments per 1,000 sessions
Organic search4,326296.7
Direct8,339293.5
Organic social7,724172.2
Referral636711.0
Paid search93566.4
Cross-network2,18320.9
AI Assistant63—12 form submissions

Australian sessions only, so Meta's data-centre traffic is stripped out. Appointments are the schedule event, which reconciles with Ernesto's daily tracker. Paid search spend over the same period was roughly A$3,000 a month.

Appear versus win
MeasureValueRead
Impressions, 90 days612,410You are visible across the category
Clicks, 90 days3,7910.62% of the time you appear
Average position26.3Page three
Pages with any impression1,601 of 2,014Indexing is not the problem
Pages with any click402 of 2,01480% of the site earns nothing
Top 10 pages' share of clicks62%Everything rests on a handful of pages
Articles published127The content exists
Clicks from all 127 articles2115.5% of site clicks
Two measurement problems worth fixing in week one

Both were found while pulling your data. Neither is what you would hire an agency for, and both are costing you money right now. They are also why Measurement scores 5 out of 25.

1. Your analytics are counting robots as customers. Of the 60,589 sessions in your GA4 over the last 90 days, 36,020 came from facebook.com referrals landing in Prineville, Luleå, Forest City, Dublin and Altoona. Those are Meta's data centres, not people. Your single biggest channel by volume is bot traffic. Real Australian sessions are 24,214, not 60,000. Every report you have read, and every judgement about what is working, has been made on numbers roughly 60% inflated. GA4
2. Google Ads has been optimising against a broken signal. Your ads_conversion_Book_appointment_1 event fired 1,079 times in 90 days. Your real booking event fired 91 times in the same window, which matches the 38 a month in Ernesto's own tracker once phone and walk-ins are added. The conversion you are feeding Google is out by roughly twelve times. Your own exports show the damage: 485 conversions recorded in Q1 2026, then 11 in Q2. Google has been buying the wrong clicks for months because it was told the wrong thing was working. GA4 Google Ads

Appendix B: the scorecard in detail

PillarLevelWhat we found
Technical
Can AI read you?
3 of 5 · 15 pts Good bones. Valid Product and Offer schema on product pages, Organization, LocalBusiness and BreadcrumbList sitewide, clean sitemaps, HTTPS, and a fast server at 80ms. Three things hold you at 3. Your robots.txt blocks GPTBot while welcoming every other AI crawler, so the single most important one is shut out. Your llms.txt is written in robots.txt syntax, which does nothing, and it contradicts the real file by saying GPTBot is allowed. And mobile Core Web Vitals fail badly: Lighthouse 59 with an LCP of 14.0 seconds. Level 4 requires Core Web Vitals to pass.
Content
Do you answer the questions?
2 of 5 · 10 pts Plenty of volume, almost no effect. 2,014 pages and 127 articles, but 80% of pages earned zero clicks in 90 days and a single one of your 112 target terms sits in the top ten. There is no FAQ schema anywhere on the site, which is the clearest AEO signal, and no answer-first structure for AI to lift from. Whole territories are uncovered: three pages for diamond education, none at all for remote and interstate buying.
Authority
Do engines trust you?
3 of 5 · 15 pts Stronger than most at your size. DR 38, 151 live referring domains, and genuine editorial coverage including Marie Claire Australia, Man of Many and Word of Mouth. The profile is unmanaged though, with a visible tail of press-release syndication, bookmark directories and a few links from piracy sites that add nothing. There is also no Review or AggregateRating schema anywhere, so the trust you have earned in person is invisible to search and to AI.
Measurement
Can you see it working?
1 of 5 · 5 pts The two problems in Appendix A sit here: a booking conversion out by twelve times, a paid channel optimising against it, and 60% of reported traffic coming from Meta's data centres. There is no organic channel reporting and no AI visibility tracking of any kind. Analytics being installed is table stakes and does not lift this off level 1.

Appendix C: the demand detail

TerritoryTermsSearches / moPromptsPages liveTerms rankingAvg position
Engagement rings (core)1951,60511374525.9
Tennis bracelets & fine jewellery1146,74010568618.2
Wedding rings & bands1434,89012163437.5
Diamond shapes2426,210232341528.8
Lab grown1324,8751537647.4
Coloured stones98,820815520.3
Diamond education84,7301530—
Metals & settings44,190104141.6
Local Sydney51,590120—
Religious jewellery343051850—
Trust & aftercare0—96——
Custom & bespoke process230770—
National & remote0—100——
Total112204,1101361,59842—

Across 1,598 pages and 204,110 searches a month of demand, exactly one term sits in the top ten. Territories with no keyword volume are pure AI-answer plays, where people ask assistants questions they never typed into Google.

Appendix D: where to play, your content territories

TerritoryMoveWhy, and what it takes
Lab grownBuild24,875 searches a month, the fastest-growing demand in the category, and you have 37 pages sitting at an average position of 47. The interstate signal is loudest here too: Australians search "lab grown diamond rings" city by city, Brisbane, Adelaide, Perth, Canberra, Melbourne, Wollongong, Cairns, Hobart, Newcastle and Geelong. This one territory is both your biggest content gap and your clearest route out of Sydney.
Engagement ringsFix51,605 searches a month and 374 pages already live. Nothing needs writing here. The category and collection pages need rebuilding so they can hold a top-ten position.
Diamond shapesFixYour best-covered territory with 234 pages and 15 terms ranking, which proves the model works when the pages exist. It is still at an average position of 28.8, so the same rebuild applies.
Diamond educationBuildThree pages against 4,730 searches and 15 buyer questions. This is where AI answers get sourced, and where a master jeweller should be the obvious authority.
Trust & aftercareBuildAlmost no Google volume, but nine high-value questions people ask assistants: resetting an inherited diamond, resizing, insurance, servicing. Pure AEO, and a genuine differentiator for a workshop that actually does this.
National & remoteBuildZero pages today. This is the Zoom business in content form: how a remote consultation works, how a ring gets designed over video, how it ships safely, what happens if the size is wrong.
Religious jewelleryHold185 pages against 430 searches a month. Small, but you own a real niche here and your Orthodox pages already earn clicks. Worth keeping tidy, not worth investing in.

Appendix E: who you are up against, and the authority play

BrandDROrganic traffic / moRanking keywordsKeywords in top 3
Ernesto Buono381,34051639
Cerrone—10,4421,760443
Larsen Jewellery4830,5003,2191,496
Temple & Grace4440,4956,8461,252

Ahrefs, Australian market, 4 August 2026. Ahrefs estimates your organic traffic at 1,340 a month and your Search Console records about 1,264, which is close enough to trust the comparison.

The important read. Larsen sits ten Domain Rating points above you and converts that into 38 times as many top-three rankings. Temple & Grace sits six points above you with 30 times your traffic. If this were an authority gap it would take years and a large budget to close. It is not. They have simply built pages that can rank, and you have not. That is what makes this a six-month engagement rather than a three-year one.

Appendix F: where the numbers come from

What is validated and what is modelled. Everything about where you sit today is measured: impressions, clicks, positions, sessions, appointments, authority and technical state all come from your own accounts or from live tools. Everything about where you could get to is modelled, and rests on the share of the pool you win, which is the one number nobody can promise. We have shown three cases so you can judge it yourself.