One buyer question per brand
The question a customer would actually type, not a keyword. Tracked as a topic, unchanged for the whole run.
Case studies
AI visibility case studies on real brands
We pick a household name, a question its buyers ask, and three AI assistants — then check every day for a week and publish everything that came back.
The same thing happened to every brand here, and it is the useful part. All four were named in all 21 of their answers — Amazon, Shopify, HubSpot and Apple, no exceptions. Being mentioned is where a household name starts, not where it competes. What separates them is everything after that: Shopify and HubSpot came first in every single answer, Amazon slipped to second inside one assistant, and Apple lost first place in seven answers out of 21. Share of voice ran 19%, 23%, 23% and 30% — not one of them held even a third of the conversation, because the assistants named anywhere from six to thirteen rivals beside them. A brand that stops at “are we mentioned?” records four flat, perfect weeks and learns nothing from any of them.
14 of 21answers that actually put Apple in first placeApple in AI Answers: First in 14 of 21
Seven runs, three AI assistants, 21 answers to one buyer question. Apple was named every time — and two of the three assistants still put Samsung above it.
Read the study →
8 of 13rivals that only one of the three assistants ever namedHubSpot in AI Answers: Three Rival Lists
Seven days, three AI assistants, 21 answers to one CRM question. HubSpot led every time — and 8 of its 13 rivals were named by just one assistant.
Read the study →
72 vs 70Walmart outscored Amazon across Gemini's seven answersAmazon vs Walmart in AI Answers: A 7-Day Study
We put one shopping question to ChatGPT, Gemini and Claude every day for a week. Amazon was named in all 21 answers — Gemini still scored Walmart higher.
Read the study →
23%share of voice, after finishing first in all 21 answersShopify in AI Answers: First, but 23% Share
Seven days, three AI assistants, 21 answers. Shopify was named first in every one and still held just 23% share of voice. The full dataset, and why.
Read the study →Method
How every study on this page is run
The same four rules each time, so two studies can be compared with each other and not just read one at a time.
Three assistants, every day
ChatGPT, Gemini and Claude, each asked the raw question with no system prompt, each pinned to the tier a free-plan user really gets.
One judge, one scale
Every answer from every assistant is scored by the same model, so a Gemini number and a ChatGPT number belong in the same column.
Nothing illustrative
Every figure is recomputed from the stored answers by the same function that draws the dashboard. The screenshots are the real thing.
Run the same study
on your own brand.
One free check, no account — or track every buyer question daily across all three assistants.