Case study
Elite Slabs
How AgentLocal measured and improved AI visibility for a stone benchtop fabricator in Beaconsfield, Victoria — with defined queries, honest baselines, and re-tested results.
Reviewed
The business
Elite Slabs Pty Ltd is a stone benchtop fabricator and installer based in Beaconsfield, Victoria (ABN 82 613 142 135, registered since June 2016). They fabricate and install natural stone, porcelain and silica-free mineral benchtops for kitchens, bathrooms and commercial projects across Melbourne’s south-east.
Like many trade businesses, their online presence was thin: a basic website with no structured data, an unclaimed Google Business Profile with outdated hours, and no structured service information that AI systems could parse.
The measurement setup
We defined 50 customer-style queries representing real searches people ask AI assistants when looking for benchtop fabrication in Melbourne’s south-east. These covered service queries (“stone benchtop installation Berwick”), comparison queries (“porcelain vs natural stone benchtops”), price queries, location queries, and brand queries.
- Queries defined
- 50 customer-style queries across service, comparison, price, location and brand intent.
- AI surfaces tested
- ChatGPT (with search) and Claude, using their respective APIs with web search enabled.
- Baseline period
- Two weeks of repeated runs before any changes were made.
- Competitors tracked
- Four other benchtop fabricators and suppliers serving Melbourne’s south-east.
Baseline results (before any changes)
The baseline showed that Elite Slabs was largely invisible to AI assistants. When people asked about benchtop fabrication in the Beaconsfield / Berwick / south-east Melbourne area, competitors with better-structured information were mentioned instead.
All rates show the count behind them. Baseline queries were run between 14–27 June 2026 across ChatGPT and Claude.
Information gaps identified
Comparing what AI systems could find about Elite Slabs against the queries consumers actually ask, we identified these gaps — ranked by how often they affected real queries:
- No structured service data. The website had no schema markup. AI systems couldn’t determine what services Elite Slabs offers, what materials they work with, or what areas they serve.
- Unclaimed Google Business Profile. Hours were wrong, no service categories were set, and there was no link to the website.
- No content answering common questions. Queries like “how much do stone benchtops cost in Melbourne” and “porcelain vs granite benchtops” had no Elite Slabs content to draw from.
- Inconsistent business details. The phone number on the website didn’t match the ABN listing. The address format varied across sources.
- No LLM-readable page. No llms.txt, no structured about page, no machine-readable service catalogue.
Changes made
Each change was recorded as a dated, auditable intervention. The work was performed by AgentLocal, not left to the business to figure out.
Week 1–2: Structured data and source fixes
- Added LocalBusiness + Service schema markup to the website, covering business name, address, phone, hours, services offered, materials, and service area.
- Claimed and corrected the Google Business Profile — updated hours, added service categories, linked the website, added photos of completed work.
- Fixed the phone number inconsistency between the website and ABN listing.
- Standardised the address format across all public sources.
Week 3–4: Content creation
- Published a service page answering “how much do stone benchtops cost in Melbourne” with actual pricing ranges, material comparisons, and photos.
- Published a comparison page for “porcelain vs natural stone benchtops” covering durability, cost, maintenance and aesthetics.
- Created a materials page listing every stone and porcelain type Elite Slabs works with, with descriptions and care instructions.
Week 5: AgentLocal profile and AI-readable publishing
- Created the AgentLocal business profile with verified, sourced facts — every detail linked to its public source and date checked.
- Published an llms.txt file and an AI-readable markdown version of the profile.
- Added the business to the AgentLocal API and MCP endpoint so agents can query it programmatically.
Results after 30 days
The same 50 queries were re-run on the same AI surfaces. The baseline portfolio was frozen — no queries were added or removed between measurements.
30-day re-test ran 28 July – 10 August 2026. Same queries, same surfaces, same methodology. Rates include denominators; a mention does not prove causation.
What actually changed for the business
Beyond the numbers, the practical outcome is that AI assistants can now answer questions about Elite Slabs accurately. When someone asks ChatGPT “who does stone benchtops near Berwick?” they get a correct, sourced answer instead of silence or a competitor.
The website now has structured data that any AI system can parse. The Google Business Profile is claimed and accurate. Common customer questions have dedicated, well-sourced answers. And the AgentLocal profile provides a machine-readable layer that any assistant, API client or MCP tool can access.
Important context
- These results describe this test set, time window and AI surfaces only. They do not guarantee how any assistant will answer in the future.
- A mention or citation does not prove that AgentLocal caused the recommendation. AI systems change their retrieval and ranking methods, and identical questions can produce different answers.
- We measure repeated observations across a frozen query portfolio. The improvement correlates with the changes we made; we cannot prove causation.
- Competitor data is included for context. Competitors were not notified or involved in this measurement.