Google’s AI Cited 108 Publishers About Supplements. The Same Ten Appeared in 95% of the Answers.

Share of AI answers, by brand

How often each brand was named in the AI Overview text, across the 39 answers:

The ordering is the finding. Transparent Labs and Nutricost are direct-to-consumer brands with a fraction of the shelf presence of Monster or Red Bull, and they are named three to twelve times more often. Celsius — a multi-billion-dollar public company — appears in fewer than one answer in five. Share of the AI answer tracks review-site coverage, not market share.

The bottleneck: who Google actually reads

Across the 39 answers Google cited 108 distinct editorial domains — but 81 of them appeared exactly once. Here is the share of answers citing each of the top ten:

37 of 39 answers cited at least one of those ten. Three of them — Forbes, Healthline and Garage Gym Reviews — carried 77% of the answers between them. A single specialist site, Garage Gym Reviews, was cited in 15 different answers spanning all four product categories.

Brand-owned domains accounted for 5.1% of all citation slots (15 of 294). Whatever a supplement brand publishes on its own website is, in this category, close to irrelevant to whether Google names it.

Each category has its own gatekeeper

The bottleneck is not one publication. It is a different publication per aisle, which is why a single PR relationship does not fix this:

Pre-workout — Forbes (8 of 10 answers), GQ and Fortune (5 each). Transparent Labs is named in 9 of the 10.

Creatine — Healthline and Forbes (8 each), U.S. News Health (7). Optimum Nutrition appears in all 9 answers; Nutricost in 8.

Protein — Forbes (6), Healthline (5). Optimum Nutrition and Transparent Labs tie at 7 each.

Energy drinks — the most fragmented aisle, and the only one with no dominant publisher: Garage Gym Reviews and Cozymeal (4 each), Tiger Fitness and Wired (3 each). It is also the aisle where a challenger brand has the clearest opening.

What this means

The instinct when a brand is missing from AI answers is to go fix the website — add schema, write an llms.txt, deepen the product pages. Those are worth doing. They are not what is deciding this. In this category, brand-owned pages supplied one citation in twenty; the answer was assembled almost entirely from third-party editorial coverage.

Which reframes the work. Getting named in Google’s AI answer for “best pre workout” is not primarily a technical SEO problem. It is a placement problem — the same problem trade PR has always solved, except the audience is now a language model reading a roundup, and the roundup’s ranking criteria are published on the page. Garage Gym Reviews, for instance, scores every product it covers against eight named criteria, one of which is third-party testing. That is a specification a brand can actually meet, and several brands in this study hold the relevant certifications already without appearing in the roundups that score for them.

The second implication is about who is exposed. A brand with heavy retail distribution and light editorial coverage looks strong in every dashboard it owns and is nearly absent from the surface where a new customer asks what to buy. That gap does not show up in traffic reports, because there was never a session to report.

If you want the practical version of this for your own category, that is what our AI Search Visibility guide covers, and the AI Visibility Index tracks the same question monthly for local businesses.

Methodology

Data collected 2 August 2026, United States, English, desktop. Forty buying-intent queries were distributed evenly across four clusters — energy drinks, pre-workout, creatine and protein — and submitted to Google via the DataForSEO live SERP endpoint with AI Overview loading enabled. Thirty-nine returned an AI Overview; the one that did not (“creatine monohydrate brand recommendations”) is excluded from all percentages rather than counted as a miss.

For each answer we recorded the full AI Overview body text and every source domain in its reference block. Thirty brands were scored for appearance in the body text using word-boundary matching on a curated list of name variants per brand; short or ambiguous names (Bang, Reign, Bloom, RAW) required the full brand phrase, so that ordinary prose could not manufacture a mention. “Editorial” publisher counts exclude brand-owned domains and platform surfaces (Google, Amazon, Reddit, YouTube, Instagram, Facebook and retailer domains), which are reported separately: Reddit was cited in 14 of 39 answers and Amazon in 14.

Four limits worth stating plainly. This is a single snapshot, not a trend — AI Overviews are non-deterministic and vary by user, location and time, so individual query results will not always reproduce. It measures Google’s AI Overview specifically, which is a different surface from Google AI Mode, ChatGPT or Perplexity, and results should not be generalized across them. A brand referred to only by product name, without the brand name attached, would be scored as absent. And appearing in an AI answer is a visibility measure, not a revenue measure — we make no claim here about traffic or sales impact.

Underlying data, including the full query list and per-query citation records, is available on request under CC-BY.