How do you measure "share of voice" in AI answers?

We run a fixed set of 30 category buy-intent prompts through ChatGPT and Gemini every week and log each brand mention and citation against the specific prompt, engine and week on a dashboard we built. Share of voice is one brand's mentions as a percentage of all brand mentions across that set. The number in this case study is the 2026-W37 snapshot.

Why would a B2B dental-materials distributor care about AI answers?

Because buyers increasingly ask an assistant "where do I buy reputable dental materials" before they open Google, and act on the names that come back. If no assistant names you at that step, you're out of consideration before the buyer reaches your site — and it never shows in web analytics, because the visit never happens.

Is this the same as SEO ranking?

No. This measures presence and share inside AI-generated answers, not blue-link position in search results. The two can move independently, so we report the AI-answer number on its own and don't fold search-ranking figures into it.

Was Oroca named with consent, and does this include sales figures?

Oroca is named with client approval. The dashboard measures presence in AI answers only — we don't attribute revenue or orders to it, and a past measurement for one brand isn't a promise of the same result for another.

Core delivery work·AVOS sits on top →

CASE · AI VISIBILITY · SHARE OF VOICE

Built AI-visible from day one.
0 → 45% share of voice.

Oroca distributes dental materials to clinics and labs across Ho Chi Minh City and Hanoi. Its site was a brand-new build — and we designed it from day one to be read correctly by the AI assistants its buyers now ask for a supplier before they open Google. Today Oroca holds 45.0% share of voice in its category's AI answers: the outright #1, more than the next three competitors combined and about 3.3 times the nearest rival — up from zero at launch, and re-measured every week.

Client

Oroca · oroca.vn

Sector

Dental-materials distribution · B2B · Vietnam

Engagement

New build, designed for AI visibility + weekly measurement

Measured

Weekly · snapshot 2026-W37

Status

● Live

01 · The context

Buyers ask an AI before they ask Google.

Oroca supplies B2B dental materials to clinics and labs in Ho Chi Minh City and Hanoi. More and more of its buyers open an AI assistant first — "where do I buy reputable dental materials in Vietnam?", "authorized distributor in Hanoi" — and act on the names that come back.

If an assistant doesn't mention you at that moment, you're out of consideration before the buyer ever reaches your site. Because Oroca's site was a brand-new build, we didn't have to retrofit fixes onto a legacy site — we could design it to be read correctly by AI from the very first day.

02 · Before

A new site, not yet in the answer.

At launch, a brand-new site has no presence in AI answers by default — the assistants haven't yet resolved who it is or what it sells. Ask one for a reputable dental-materials supplier in Vietnam and it named competitors, not Oroca. Share of voice started at zero.

There was also no way to see it. Nobody was measuring which brands the assistants named, so the gap was invisible as well as unaddressed.

03 · What we did

Build it AI-readable from day one — and measure it.

Because this was a new build, we could do two things properly from the start rather than patching them on later: make the whole store machine-readable for AI, and stand up the measurement to prove where it stood.

Built the entire ecommerce site to be machine-readable for AI — clean, consistent, structured product data and business information, so assistants resolve exactly who Oroca is, what it sells and who it serves.
Built a realtime visibility-monitoring dashboard — 30 category prompts run weekly across ChatGPT and Gemini, tracking share of voice and citation rate by prompt, engine and week, updated in realtime so the client can click in and verify any time.

04 · How we measured

Every mention tied to a prompt, an engine and a week.

Each week the same 30 category prompts run through ChatGPT and Gemini, and every brand mention and citation is logged against the exact prompt, engine and week on the dashboard. Share of voice is Oroca's mentions as a percentage of all brand mentions across that set.

The figures below are the 2026-W37 snapshot. Because the set runs on a schedule rather than once, this is a verifiable trend line the client can click into — not a single screenshot.

05 · The result

#1 share of voice — and still climbing.

In the 2026-W37 snapshot Oroca holds 45.0% share of voice — 49 of the category's brand mentions — the outright #1. That is about 3.3 times the nearest rival (Dental Shop, 13.8%) and more mentions than the next three competitors combined (49 versus 39). The remaining five competitors share the other 55.0%.

And it's a rising line, not a one-off: zero at launch, 41.2% in W36, 45.0% in W37 — measured weekly on the same prompt set. Across the 30-prompt set Oroca's AI mention rate is 82% — 27 cited and 22 mentioned — and it holds on both engines: ChatGPT 76.7%, Gemini 86.7%.

Oroca — 49 mentions · 45.0%
Dental Shop — 15 · 13.8%
Dental Floor — 13 · 11.9%
Minh Khang Dental — 11 · 10.1%
Nha Viet Dental — 11 · 10.1%
Dentaluck — 10 · 9.2%

0 → 45.0%

Share of voice in AI answers

From zero at launch to the outright #1 in its category — 49 brand mentions across ChatGPT and Gemini, tracked weekly (0 → 41.2% → 45.0%).

82%

AI mention rate

27 cited and 22 mentioned across the 30-prompt set. ChatGPT 76.7%, Gemini 86.7%.

~3.3×

Ahead of the nearest rival

More mentions than the next three competitors combined; the other five share the remaining 55.0%.

IS AN AI NAMING YOU?

See your share of voice
in AI answers.

The same weekly measurement we run for Oroca — 30 category prompts across ChatGPT and Gemini, on a dashboard you can click into.

Check my AI visibility See all work

Built AI-visible from day one.0 → 45% share of voice.