Census · Published 2026-10-22 · recorded, not live

AI named a brand in 47% of answers. Named most: Nature Made, in 30%.

What a leading AI assistant names when a shopper asks about multivitamins, from a budget pick to one for men over 50. The first census for this category, 64 answers. 6,000+ data points measured by the Supracited AI engine. The buyer scenarios are further down.

Supracited AI engine

Six buyer scenarios for multivitamins, each compartment one buyer type, filled with one pill per recorded answer, coloured by the brand its answer named first. The first census for this category, 64 answers in all.

Ritual was named first most often, in 14% of answers, more than any other tracked brand.

Each compartment, its base, the brand named first most and how often no tracked brand was named first, with one buyer question it answered
CompartmentAnswersNamed first mostCountNo brand firstA real buyer question
Best overall8No brand0 of 88 of 8What is the best time of day to take a multivitamin?
Budget8Nature Made6 of 80 of 8I'm 22 and on a student budget, what's a solid multivitamin that won't break the bank?
Gummy or capsule8No brand0 of 88 of 8are gummy vitamins better than capsules
What is folate8No brand0 of 88 of 8What is folate and what does it do?
Combining with meds8No brand0 of 88 of 8I am taking medications (the blood thinner Coumadin and an antidepressant) and several supplements (green tea extract, CoQ10, and St. John's wort). Is it OK to take all of these together?
Men over 508Centrum5 of 80 of 8What multivitamins are recommended for men over 50?

  • 9%

    Named a store brand

    Kirkland Signature, the store named most.

  • 5 of 8

    Led the men-over-50 pick

    Centrum, the brand named first most for that scenario.

  • 22%

    Cited healthline.com

    The site cited most.

Who answered: A leading AI assistant with web search on, 64 answers, recorded 2026-10-22. Assistants are identified to clients in their reports. Brand names here are measurement results, not recommendations.

Every answer, by buyer scenario

Supracited AI engine

Each dot is one recorded AI answer, collected by the Supracited AI engine: 8 for each buyer scenario, 64 in all. Pick a brand to see which answers named it.
Brand
  1. Best time to take0 of 80 of 80 of 80 of 80 of 8
  2. Budget8 of 80 of 86 of 80 of 80 of 8
  3. Gummy vs. capsule0 of 80 of 80 of 80 of 80 of 8
  4. What is folate0 of 80 of 80 of 80 of 80 of 8
  5. Combining with meds0 of 80 of 80 of 80 of 80 of 8
  6. Other scenario A5 of 86 of 80 of 80 of 80 of 8
  7. Other scenario B6 of 87 of 80 of 88 of 80 of 8
  8. Men over 500 of 86 of 86 of 80 of 86 of 8

Nature Made

30%

Ritual

30%

Centrum

19%

Olly

13%

Garden of Life

9%

Nature Made: named in every one of them on a budget multivitamin, against none of them on five buyer scenarios each.

Ritual: named in some of them on a scenario not shown individually, against none of them on five buyer scenarios each.

Centrum: named in some of them on a budget multivitamin and a multivitamin for men over 50 each, against none of them on six buyer scenarios each.

Olly: named in every one of them on a scenario not shown individually, against none of them on seven buyer scenarios each.

Garden of Life: named in some of them on a multivitamin for men over 50, against none of them on seven buyer scenarios each.

Each card shows the brands an answer named, in the order it named them. We quote only sentences we have checked as plain fact, such as what a brand is or where it is sold, and none of these answers has one.

Answers naming each brand, by buyer scenario, out of 8 each
Buyer scenarioNature MadeRitualCentrumOllyGarden of Life
Best time to take0 of 80 of 80 of 80 of 80 of 8
Budget8 of 80 of 86 of 80 of 80 of 8
Gummy vs. capsule0 of 80 of 80 of 80 of 80 of 8
What is folate0 of 80 of 80 of 80 of 80 of 8
Combining with meds0 of 80 of 80 of 80 of 80 of 8
Other scenario A5 of 86 of 80 of 80 of 80 of 8
Other scenario B6 of 87 of 80 of 88 of 80 of 8
Men over 500 of 86 of 86 of 80 of 86 of 8

The names the answers gave

Each bar is the share of recorded answers that named it. Hover or tap a name for its count, its range and what the count means.

64 recorded answers, eight buyer scenarios each asked 8 times, collected 2026-10-22.

  1. Named in 64% of recorded answers (41 of 64). Range if the census ran again: 34 to 91%. Effective n: 11 of 64 answers (ICC 0.67, design effect 5.7). In the first three names: 36 of 64. Counts mentions, including cautions; not recommendations.
  2. Named in 44% of recorded answers (28 of 64). Range if the census ran again: 17 to 72%. Effective n: 12 of 64 answers (ICC 0.62, design effect 5.3). In the first three names: 23 of 64. Counts mentions, including cautions; not recommendations.
  3. Named in 39% of recorded answers (25 of 64). Range if the census ran again: 13 to 69%. Effective n: 11 of 64 answers (ICC 0.67, design effect 5.7). In the first three names: 22 of 64. Counts mentions, including cautions; not recommendations.
  4. Named in 34% of recorded answers (22 of 64). Range if the census ran again: 9 to 63%. Effective n: 13 of 64 answers (ICC 0.58, design effect 5.1). In the first three names: 11 of 64. Counts mentions, including cautions; not recommendations.
  5. Named in 30% of recorded answers (19 of 64). Range if the census ran again: 5 to 59%. Effective n: 10 of 64 answers (ICC 0.74, design effect 6.2). In the first three names: 6 of 64. Counts mentions, including cautions; not recommendations.
  6. Named in 30% of recorded answers (19 of 64). Range if the census ran again: 5 to 58%. Effective n: 11 of 64 answers (ICC 0.70, design effect 5.9). In the first three names: 3 of 64. Counts mentions, including cautions; not recommendations.
  7. Named in 19% of recorded answers (12 of 64). Range if the census ran again: 0 to 44%. Effective n: 11 of 64 answers (ICC 0.68, design effect 5.8). In the first three names: 4 of 64. Counts mentions, including cautions; not recommendations.
  8. Named in 17% of recorded answers (11 of 64). Range if the census ran again: 0 to 44%. Effective n: 10 of 64 answers (ICC 0.79, design effect 6.5). In the first three names: 5 of 64. Counts mentions, including cautions; not recommendations.
  9. Named in 13% of recorded answers (8 of 64). Range if the census ran again: 0 to 38%. Effective n: 8 of 64 answers (ICC 1.00, design effect 8.0). In the first three names: 1 of 64. Counts mentions, including cautions; not recommendations.
  10. Named in 9% of recorded answers (6 of 64). Range if the census ran again: 0 to 30%. Effective n: 11 of 64 answers (ICC 0.71, design effect 6.0). In the first three names: 6 of 64. Counts mentions, including cautions; not recommendations.
  11. Named in 9% of recorded answers (6 of 64). Range if the census ran again: 0 to 31%. Effective n: 11 of 64 answers (ICC 0.71, design effect 6.0). In the first three names: 0 of 64. Counts mentions, including cautions; not recommendations.
Show 6 more names
  1. Named in 8% of recorded answers (5 of 64). Range if the census ran again: 0 to 20%. Effective n: 27 of 64 answers (ICC 0.19, design effect 2.4). In the first three names: 1 of 64. Counts mentions, including cautions; not recommendations.
  2. Named in 6% of recorded answers (4 of 64). Range if the census ran again: 0 to 20%. Effective n: 16 of 64 answers (ICC 0.43, design effect 4.0). In the first three names: 2 of 64. Counts mentions, including cautions; not recommendations.
  3. Named in 6% of recorded answers (4 of 64). Range if the census ran again: 0 to 20%. Effective n: 16 of 64 answers (ICC 0.43, design effect 4.0). In the first three names: 3 of 64. Counts mentions, including cautions; not recommendations.
  4. Named in 3% of recorded answers (2 of 64). Range if the census ran again: 0 to 13%. Effective n: 32 of 64 answers (ICC 0.14, design effect 2.0). In the first three names: 1 of 64. Counts mentions, including cautions; not recommendations.
  5. Named in 3% of recorded answers (2 of 64). Range if the census ran again: 0 to 13%. Effective n: 32 of 64 answers (ICC 0.14, design effect 2.0). In the first three names: 1 of 64. Counts mentions, including cautions; not recommendations.
  6. Named in 3% of recorded answers (2 of 64). Range if the census ran again: 0 to 11%. Effective n: 32 of 64 answers (ICC 0.14, design effect 2.0). In the first three names: 0 of 64. Counts mentions, including cautions; not recommendations.
  • Share of answers that named it
  • Range if the census ran again

Asked through the Supracited AI engine, with web search on, rather than inside the assistant’s own app, which may answer differently.

See the numbers as a tableevery name given, with both ranges, position and top three

Each row leads with a cluster-bootstrap range, which allows for the same question being asked again; the narrower Wilson range beside it assumes every answer is an independent draw. The method page explains why the bootstrap leads.

Exploratory: 8 buyer scenarios. An industry framework (IAB, August 2026) treats a program with fewer than 50 buyer scenarios as exploratory rather than directional; read these ranges as a guide, not a precise estimate. Known limits.

ICC, the intraclass correlation, is how alike the answers to one buyer scenario are: 0 is no more alike than chance, 1 is the same answer every time. The design effect is how much that widens a range, and effective n is the answers divided by it, roughly what they are worth as separate draws.

64 answers with web search

64 answers: 8 buyer scenarios, each asked 8 times, collected 2026-10-22. Asked through the Supracited AI engine, with web search on, rather than inside the assistant’s own app, which may answer differently.

Narrow screen: scroll the table sideways for more columns.

n=64 answers · 8 buyer scenarios, each asked 8 times · collected 2026-10-22
Name Kind Share of answers (bar: bootstrap range, tick: observed) Wilson range (assumes independent answers) ICC · design effect · effective n Average position when named (1 = first) In the first three names
IronIngredient

34 to 91% · 41 of 64

52 to 75%0.67 · 5.7 · 111.8036 of 64 (56%)
Vitamin B12Ingredient

17 to 72% · 28 of 64

32 to 56%0.62 · 5.3 · 122.8223 of 64 (36%)
Vitamin DIngredient

13 to 69% · 25 of 64

28 to 51%0.67 · 5.7 · 112.0822 of 64 (34%)
FolateIngredient

9 to 63% · 22 of 64

24 to 47%0.58 · 5.1 · 133.0011 of 64 (17%)
Nature MadeBrand

5 to 59% · 19 of 64

20 to 42%0.74 · 6.2 · 104.426 of 64 (9%)
RitualBrand

5 to 58% · 19 of 64

20 to 42%0.70 · 5.9 · 115.263 of 64 (5%)
CentrumBrand

0 to 44% · 12 of 64

11 to 30%0.68 · 5.8 · 114.004 of 64 (6%)
MagnesiumIngredient

0 to 44% · 11 of 64

10 to 28%0.79 · 6.5 · 103.455 of 64 (8%)
OllyBrand

0 to 38% · 8 of 64

7 to 23%1.00 · 8.0 · 86.001 of 64 (2%)
Kirkland SignatureStore brand

0 to 30% · 6 of 64

4 to 19%0.71 · 6.0 · 111.176 of 64 (9%)
Garden of LifeBrand

0 to 31% · 6 of 64

4 to 19%0.71 · 6.0 · 115.670 of 64 (0%)
One A DayBrand

0 to 20% · 5 of 64

3 to 17%0.19 · 2.4 · 275.601 of 64 (2%)
SmartyPantsBrand

0 to 20% · 4 of 64

3 to 15%0.43 · 4.0 · 163.252 of 64 (3%)
CVS HealthStore brand

0 to 20% · 4 of 64

3 to 15%0.43 · 4.0 · 163.253 of 64 (5%)
EquateStore brand

0 to 13% · 2 of 64

1 to 11%0.14 · 2.0 · 323.001 of 64 (2%)
MegaFoodBrand

0 to 13% · 2 of 64

1 to 11%0.14 · 2.0 · 324.001 of 64 (2%)
WalgreensStore brand

0 to 11% · 2 of 64

1 to 11%0.14 · 2.0 · 325.000 of 64 (0%)
Up&UpStore brand

0% · 0 of 64

0 to 6%Not namedNot named0 of 64 (0%)

Store brands are counted by name here. Reading each mention for whether it names a product to buy, as the pain relief census did for its main table, is still to come.

The sites the answers were built from

The twelve domains cited most, ranked by how many answers linked them. Hover or tap a site for what it is.

The twelve sites cited most, out of 64 recorded answers.

  1. Cited in 22% of recorded answers (14 of 64). What it is: Health information site. Cited in: linked at least once in the answer.
  2. Cited in 17% of recorded answers (11 of 64). What it is: Independent supplement testing service. Cited in: linked at least once in the answer.
  3. Cited in 13% of recorded answers (8 of 64). What it is: Licensed-pharmacist Q&A site. Cited in: linked at least once in the answer.
  4. Cited in 13% of recorded answers (8 of 64). What it is: UK supplement retailer. Cited in: linked at least once in the answer.
  5. Cited in 13% of recorded answers (8 of 64). What it is: US National Institutes of Health research database. Cited in: linked at least once in the answer.
  6. Cited in 13% of recorded answers (8 of 64). What it is: Biomarker testing and wellness membership service. Cited in: linked at least once in the answer.
  7. Cited in 11% of recorded answers (7 of 64). What it is: Canadian Journal of Hospital Pharmacy, a peer-reviewed journal. Cited in: linked at least once in the answer.
  8. Cited in 11% of recorded answers (7 of 64). What it is: Business magazine. Cited in: linked at least once in the answer.
  9. Cited in 11% of recorded answers (7 of 64). What it is: Parenting and family lifestyle site. Cited in: linked at least once in the answer.
  10. Cited in 11% of recorded answers (7 of 64). What it is: Ritual's own brand site. Cited in: linked at least once in the answer.
  11. Cited in 11% of recorded answers (7 of 64). What it is: Telehealth platform. Cited in: linked at least once in the answer.
  12. Cited in 9% of recorded answers (6 of 64). What it is: Independent supplement review and scoring site. Cited in: linked at least once in the answer.
  • Share of answers that linked the site at least once

Cited in: the number of the 64 answers that linked the domain at least once. Types checked on each site’s own homepage on 2026-10-22.

The buyer scenarios asked

Written the way a shopper types, naming no brand. How each one is run, counted and checked: the method page.

Hover, focus or tap a scenario to read the question as asked.

Two buyer scenarios are counted in the totals above, not shown individually.

Correctionsevery change to this page, dated
  • 2026-10-22: first publication. A pilot census: 64 answers, one wave; the base for a larger 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 we count

Other censuses

Want your own brand on this shelf?

Start with three buyer scenarios. We send back which brands AI named, which sites it cited and one source worth a closer look.

Get a free Shelf Check