CASE STUDY · GEO · ANONYMIZED · LUXURY

A Swiss luxury watchmaker, audited end to end.

180+ prompts across 9 topic clusters, 4 AI models, a year of social listening, and an action plan of 30 prioritized fixes. Brand surfacing on every branded prompt with the wrong narrative — and missing on generic categories.

180+ prompts across 9 topic clusters, 4 AI models, a year of social listening, and an action plan of 30 prioritized fixes. Brand surfacing on every branded prompt with the wrong narrative — and missing on generic categories.

Prompts › engines

Delivered

Prompt clusters

Branded

Category

Comparative

Heritage

AI engines

ChatGPT

Gemini

Claude

Perplexity

Measured

Mapped

Prioritized

Client

Anonymized — Swiss luxury watchmaker

Industry

Premium horology

Service area

SEO & GEO

Engagement type

Project (one audit cycle)

Status

Delivered

180+

180+

prompts across 9 topic clusters

4

4

AI models — ChatGPT · Gemini · Claude · Perplexity

6,000+

6,000+

mentions across a year of social listening

30

30

actions — Do Now · Do Soon · GEO-Native · Later

01 — The brief

The brief.

The brief.

The brief.

A Maison with deep heritage, strong organic SEO, and a brand the marketing team had spent years carefully building. The GEO problem was different from what most luxury brands assume.

A Maison with deep heritage, strong organic SEO, and a brand the marketing team had spent years carefully building. The GEO problem was different from what most luxury brands assume.

The brand was surfacing on every branded prompt across the major AI engines — but the narrative was wrong. AI answers were pulling from outdated sources, misattributing model history, and confusing the brand with competitors in the same horological category. On generic category prompts — the kind a high-intent buyer would type without yet knowing which Maison they want — the brand was missing entirely from some of its core categories.


The ask: measure how the brand actually shows up across the AI engines today, identify where the gaps are, and give the marketing team an action plan they can sequence and execute.

02 — What we did

What we did.

What we did.

What we did.

01

Prompt design

Built a 180+ prompt audit across nine topic clusters: branded queries (the brand by name), category queries (horological style and price tier), comparative queries (head-to-head against named competitors), heritage queries (model history, provenance, craftsmanship), and several adjacent buyer-intent clusters. The set was designed to mirror how real buyers ask AI about the category, not to maximize coverage on synthetic terms.

02

Multi-engine measurement

Every prompt run across ChatGPT, Gemini, Claude, and Perplexity. Same prompt, four engines, four answer patterns. Brand presence scored at the engine level because each engine surfaces different sources and reaches different conclusions.

03

Brand visibility scoring

For every prompt where the brand could plausibly appear, we measured whether it did — and how. Mentioned vs. recommended. Cited from the brand’s own content vs. paraphrased from someone else’s. Framed accurately vs. confused with a competitor. Every gap traced back to a content, structure, or authority root cause.

04

Social listening overlay

A year of social listening data across Reddit, Quora, and category forums — 6,000+ mentions of the brand and its competitors. Community narrative leaks into AI training corpora at much higher rates than brand-owned content; mapping that narrative was essential to understanding why the AI answers were what they were.

05

Action plan

30 prioritized actions across four time horizons — Do Now (immediate fixes the brand owns), Do Soon (structural changes in the next quarter), GEO-Native (new content built specifically for AI retrieval), Later (long-horizon authority work). Each action mapped to a specific gap, with the team that owns it named explicitly.

03 — How the engagement runs

How the engagement runs.

How the engagement runs.

How the engagement runs.

Fixed-price project across an eight-week audit cycle. Delivered as a structured action plan plus the underlying data, plus working sessions with the marketing team to align on sequencing and ownership. The audit itself was the deliverable; execution sat with the brand’s internal team and existing content partners.

04 — What changed

What changed.

What changed.

What changed.

The brand now has a baseline.

The brand now has a baseline.

Visibility across the four major AI engines is measurable. Gaps are mapped to specific content, structural, and authority root causes. The action plan is sequenced — the team knows what to do first, what waits, and what’s not worth doing at all.


The more substantive change: the marketing team has language for a problem that was previously hard to even describe. “AI is misrepresenting our heritage” became “on prompts X, Y, and Z, the brand is being paraphrased from outdated source A; we need to update content B and authority signal C.” The conversation moved from anecdotal to operational.


Future audit cycles can measure progress against this baseline. GEO is a measurable channel for this brand now, in a way it wasn’t before the audit.

Got messy AI visibility? Let’s measure it.

20-minute call. No slides. We look at what's broken and tell you what we'd do.

Got messy AI visibility? Let’s measure it.

20-minute call. No slides. We look at what's broken and tell you what we'd do.