AI Shelf Index · Supplements
Which supplement brands does AI recommend?
Every month we ask ChatGPT, Perplexity and Gemini the same 3 supplements questions with web search on, and count the brands they name. Same prompts every month, so the trend is the story.
The first supplements edition is being collected.
An edition publishes once at least 90% of the month's answers are in. Below are the exact prompts it asks and what each one needs from a product page. To track your own brand now, start a weekly Shelf Watch.
What the prompts ask for
How a supplements brand gets named
Supplement prompts name a goal (sleep), an audience (women) or a quality bar (high-quality). Assistants answer them from labels and third-party testing claims, so the facts that matter are the ones on the supplement facts panel and the certificates behind it.
Put each fact on the product page as text, in your Product structured data (as an additionalProperty where schema.org has no field for it), in your Shopify category metafields and in your Merchant Center feed's product_detail and product_highlight. Then check whether the sites assistants cite say the same thing.
| The shopper asks | The fact that answers it |
|---|---|
| “magnesium for sleep” | The exact form (glycinate, citrate, oxide), the dose per serving and servings per container, as text |
| “protein powder for women” | Protein per serving, protein source, sweeteners and allergens; who it's formulated for, if it is |
| “a high-quality fish oil” | EPA and DHA per serving, the oil's form and any third-party testing, linked to the certificate |
Panel
The 3 supplements prompts
- 1.What's a good magnesium supplement for sleep? I'm shopping online in the United States.
- 2.Which protein powder brands are best for women? I'm shopping online in the United States.
- 3.Recommend a high-quality fish oil supplement. I'm shopping online in the United States.
Fixed since the first edition. How the count works and its limits are on the Index page; journalists and researchers can find citation terms and raw data on Research.
Other categories
Nine more shelves
Also useful