Studies
Your AI shelf, measured at three depths
Each study simulates your buyers asking the leading AI assistants, at least 1,000 conversations deep, then hands you a graded fix list with the evidence behind every item.
Supracited AI engine
Each cube is one simulated buyer conversation a study of this size runs.
1,000+
simulated buyer conversations
- Free and paid versions of the leading AI assistants
2,400+
simulated buyer conversations
- Free and paid versions of the leading AI assistants
4,400+
simulated buyer conversations
- Free and paid versions of the leading AI assistants
A Brand Study: 1,000+ simulated buyer conversations.
A Category Study: 2,400+ simulated buyer conversations.
A Market Study: 4,400+ simulated buyer conversations.
| Study | Simulated buyer conversations |
|---|---|
| Brand Study | 1,000+ |
| Category Study | 2,400+ |
| Market Study | 4,400+ |
See a finished sample study: 400 conversations
6,000+
Data points measured
Recorded answers, brand mentions, source citations and hand-coded details.
798
Real buyer questions in our library
76
Distinct sources checked
Launching: Supracited’s AI engine asks leading AI assistants, including ChatGPT, Gemini, Grok and Google’s AI answers, with more being added. First Shelf Checks go out from 1 November 2026.
Three studies, one method
Every study is scoped and quoted after a short briefing call. No guarantee of results, stated in the statement of work before you sign.
-
Brand Study
1,000+
simulated buyer conversations, across free and paid versions of the leading AI assistants
- Your fix list, graded and in order, with the evidence behind each item
- Your shelf: who is named for each buyer type and assistant version, as ranges with counts
- The sources the assistants lean on in your category, each one checked
- The Reveal session and your signed Shelf Report
- The day-90 Re-check, if you want it
-
Includes the day-90 Re-check
Category Study
2,400+
simulated buyer conversations, across free and paid versions of the leading AI assistants
Everything in the Brand Study, plus
- Your shelf against your category’s rivals
- Competitor content gap, outreach kit with named authors, page briefs for five pages, 90-day roadmap
- Source checks attached to every fix
- The day-90 Re-check is included
-
Market Study
4,400+
simulated buyer conversations, across free and paid versions of the leading AI assistants
Everything in the Category Study, plus
- Location: a test of whether place changes the answer in your category, and country panels for international brands
- Product listings inside AI answers
- Two rivals in depth
- A test plan written before any change ships: matched test and control pages, the measure and the bar it must clear
- The day-90 Re-check, included
Who is asking changes the answer
From our pain relief pilot: the same five pain and headache buyer scenarios, asked as written and as four different buyers, of a leading AI assistant. Motrin was named in 87% of answers when a parent was buying for a child, and in 13% when someone 68 and on a tight budget asked.
Supracited AI engine
- Just the question84%73%89%58%40%
- 24, in a hurry87%60%87%33%2%
- Parent buying for a child89%87%84%13%18%
- Comparing carefully89%67%84%69%24%
- 68, tight budget91%13%60%47%7%
Tylenol: named in 91% of answers for someone 68 and on a tight budget, against 84% for the question alone.
Motrin: named in 87% of answers for a parent buying for a child, against 13% for someone 68 and on a tight budget.
Advil: named in 89% of answers for the question alone, against 60% for someone 68 and on a tight budget.
Aleve: named in 69% of answers for someone comparing carefully, against 13% for a parent buying for a child.
Excedrin: named in 40% of answers for the question alone, against 2% for someone 24 and in a hurry.
Quoted from a recorded AI answer. Not medical advice and not our recommendation.
Quoted from a recorded AI answer. Not medical advice and not our recommendation. We quote only sentences we have checked as plain fact, such as what a brand is or where it is sold.
| Buyer | Tylenol | Motrin | Advil | Aleve | Excedrin |
|---|---|---|---|---|---|
| Just the question | 84% | 73% | 89% | 58% | 40% |
| 24, in a hurry | 87% | 60% | 87% | 33% | 2% |
| Parent buying for a child | 89% | 87% | 84% | 13% | 18% |
| Comparing carefully | 89% | 67% | 84% | 69% | 24% |
| 68, tight budget | 91% | 13% | 60% | 47% | 7% |
Pilot, October 2026: 225 simulated buyer conversations on pain and headache. Named, not recommended: these are measurement results.
Location, said plainlywhat the pilot found about US states
The same pilot found no detectable difference between US states for headache questions, so we do not promise state differences. The Market Study tests whether location matters in your category, and runs country panels for international brands.
From a free Shelf Check to Shelf Watch
Shelf Check
Free, no card. Three buyer scenarios across the leading assistants: who is named, who is named instead of you, and one source worth a closer look.
Briefing and Study Card
A 30-minute Briefing, then one page naming your buyers, the assistants, your rivals and the date of your Reveal. You sign off the card and the buyer scenarios.
The Reveal
A recorded session for whoever signs off spend, your site owner and your agency: your shelf, why it looks that way, and the First Three fixes, each with an owner.
Fix
You leave knowing Monday’s work: the graded fix list and, in the Category and Market Studies, the outreach kit, page briefs and roadmap. Your team builds, or we guide it.
The day-90 Re-check
The same buyer scenarios, asked again, then beside now: moved, held or within noise. If an assistant changed version, we say so and never credit it to your work.
Shelf Watch
A monthly page written by a person, the trend, and a call each quarter.
What you see, and what stays in our kitchen
We publish the principles that make our numbers trustworthy. The recipe stays with us.
In every study
- Free and paid versions of the leading AI assistants, each answer labelled with the version that gave it
- Every figure as a range, printed with the count it came from
- The buyer scenarios we ran, and answers quoted word for word
- Every source the assistants cite, checked, with any flag stated in words
- Your shelf in one picture: each buyer type against each assistant version, showing where you lead, where a rival leads and where it is too close to call
Coverage grows over time. Which assistants and systems answer appears in the public census and in a study client’s own report, never as a running count on this page.
Stays in our kitchen
- The full library of buyer scenarios and their phrasings
- How simulated buyers are written
- How scenarios are assigned across buyers and assistants
- Weighting, flag thresholds and the fix playbook
After the study
For study clients who want to keep watching the shelf after the Reveal, or want a hand doing the work.
-
Shelf Watch
- Your shelf each month on the assistants your buyers use most
- A monthly page written by a person, with the trend
- A call each quarter
-
Shelf Watch with Re-check
- Every assistant version we measure, each month
- A full Re-check each quarter
- The monthly page and the quarterly call
-
3 of 3 seats open
Guided 90
Three months, three clients at a time. Your team builds. We set the order from the graded fix list, review each change before it ships, meet weekly and re-check at day 90. Because we guided the work, the Re-check says so on its first page.
See the free version of this idea in the pain relief census.
We measure and source. We do not write, edit or suggest any statement about what a product does; that stays as your label states it. Brand names in our censuses are measurement results, not recommendations.
How a study runsStart with three buyer scenarios
We send back which brands AI named, which sites it cited and one source worth a closer look.