The machines describe ten healthier beverage brands well and rank none of them in a top answer.

This is ARA's first reading of healthier beverages. Four AI models answered fifteen buyer questions about twenty brands. Each brand had sixty chances to be recommended. Nothing here is a change, because there is no earlier quarter to read against.

Five brands hold the recommendation layer. Spindrift scores 81. Olipop scores 79. Liquid Death scores 77. Poppi scores 75 and LaCroix scores 70. The other fifteen brands score 63 and below.

The gap sits in one attribute. Findability, Clarity, Sentiment and Identity Fidelity stay close from the top of the board to the bottom. AI Recommendation runs from 20 down to 0. The cohort average for that attribute is 5.3.

Thirteen of the twenty brands hold real equity that the models under-reward. Three brands are recommended above their equity. Four brands sit where reward and equity agree. The category median is 56.5. The board runs from 81 down to 47.

ABOUT THIS READING

ARA scores five attributes for each brand, each out of 20, for a total out of 100. The bands are AWESOME 83 to 100, STRONG 70 to 82, AVERAGE 56 to 69, WEAK 40 to 55, and INVISIBLE 0 to 39. This cohort fills three of the five bands. Five brands are STRONG. Five are AVERAGE. Ten are WEAK.

The five attributes are Findability, Clarity, Sentiment, Identity Fidelity and AI Recommendation. Four models answered the same fifteen buyer questions for every brand. ARA also records a position for each brand. SWELL marks real equity that the models under-reward. EBB marks a brand the models reward above its equity. ALIGNED marks agreement between the two.

Winners and losers

Spindrift leads the board at 81. All four models chose it, 28 times across its 60 moments, or 47%. It was named first in 10 moments, the joint highest first-place count in the cohort. The models name Spindrift for hot-day refresher, real fruit, simple ingredients, best taste, not too sweet, natural and real, and soda swap. They do not name it for with breakfast, family favorite, cultural footprint or sourcing transparency. Its position is ALIGNED, so the reward matches the equity ARA measures.

Olipop sits second at 79 and carries the strongest single attribute in the cohort. It scores the maximum on AI Recommendation against a cohort average of 5.3. All four models chose it, 28 times across its 60 moments, and named it first in 10. Olipop wins everyday fridge stock, with breakfast, lunch on the go, best taste, not too sweet, and soda swap. The models do not name it for family favorite, real fruit, simple ingredients, natural and real, or for kids. Its position is EBB, so the models recommend Olipop above the equity ARA measures.

Naked sits last at 47. No model chose it in any of its 60 moments. It scores zero on AI Recommendation. Its Findability is 10 and its Sentiment is 11, both near the floor of the cohort. The models do not name Naked for everyday fridge stock, with breakfast, family favorite or lunch on the go. Its position is ALIGNED, so ARA finds no buried equity here.

Milo's carries the widest split between its fundamentals and its recommendation. It scores 59 overall and zero on AI Recommendation, against a cohort average of 5.3. Its Clarity is 18 and its Sentiment is 15. Both sit in the top range of this cohort. No model chose it in any of its 60 moments. Its position is SWELL, so the models under-reward what Milo's has built.

The Leaderboard.

One hundred and fifty-three brands share the healthier beverage shelf.

Across the same fifteen questions, the four models named 312 choices. Those choices spread across 153 brands. The twenty brands ARA measured hold 28.8% of them. The largest single holding is 27 choices, or 8.7% of the shelf. No brand owns a tenth of this category.

The biggest names outside the measured set are Bubly with 10 choices and Waterloo with 9. Zevia holds 8. Vita Coco holds 6. Aura Bora, Culture Pop, Harmless Harvest and Health-Ade each hold 5. The models return these names in the same answers as the brands at the top of the board.

The tail is long. One hundred and two brands are named exactly once, and together they hold 33% of all choices. Sixteen brands are named exactly twice, and they hold 10%. Together, 118 of the 153 brands hold 43% of the shelf. Close to half of this category's recommendation volume sits with brands a model names once or twice.

15 buyer questions x 4 models / audited brands in colour, every other brand in a stepped grey. Toggle between Cohort Only and Full Shelf.

KEY FINDING 01

Five brands hold nearly all the recommendation in this category.

Spindrift and Olipop both score the maximum on AI Recommendation. Poppi scores 19. LaCroix scores 18. Liquid Death scores 13. The cohort average is 5.3.

Below those five, the number drops fast. Simply scores 6. Welch's scores 4. Uncle Matt's scores 3. Tropicana scores 2 and Natalie's scores 1. Ten brands score zero.

Ten brands reach no answer's top seven. They are Florida's Natural, Gold Peak, Good & Gather, Great Value, Kroger, Milo's, Minute Maid, Naked, Simple Truth and SunnyD. Some of them appear lower in an answer. None of them reach the place a buyer reads first.

KEY FINDING 02

Milo's earns one of the clearest descriptions in the category and zero recommendation.

Milo's scores 18 on Clarity, second only to Liquid Death at 19. Its Sentiment is 15. Its total is 59, which puts it in the middle of the board. Its AI Recommendation is zero.

The pattern repeats down the board. Uncle Matt's scores 17 on Clarity and 16 on Sentiment, and 3 on recommendation. Natalie's scores the same pair and 1 on recommendation. Welch's scores 17 on Clarity and 4 on recommendation. Florida's Natural scores 17 on Clarity and zero.

Fundamentals and recommendation work as separate systems here. A model can hold an accurate, warm account of a brand and still keep it out of the answer. Clarity and Sentiment tell a brand whether the machine understands it. Recommendation tells the brand whether the machine will say its name.

Brand Fundamentals vs Recommendation.

KEY FINDING 03

Liquid Death is the only brand the models both understand and treat as distinct.

Clarity is the strongest attribute in this category. Liquid Death scores 19. Spindrift and Milo's score 18. Uncle Matt's, Natalie's, Welch's, Tropicana, Florida's Natural, Olipop and Poppi all score 17. The models know what these brands sell.

Identity Fidelity does not follow. Liquid Death scores 17 and stands alone at the top. Simply scores 15. Spindrift, Kroger and Gold Peak score 14. Most of the cohort sits at 11, 12 or 13. The models describe these brands accurately and hold no sharp sense of who they are.

Liquid Death holds both halves. It scores 19 on Clarity and 17 on Identity Fidelity, and it ranks third overall at 77. Its AI Recommendation is 13, the lowest of the top five. The models understand Liquid Death and tell it apart from the shelf, and they still name it in only 6 of its 60 moments.

Understood vs Distinct.

KEY FINDING 04

Poppi is named often and Spindrift is named first.

Poppi appears in 25 of its 60 moments. It is named first in 1. Poppi scores 19 on AI Recommendation and ranks fourth at 75. The models name Poppi often. They rarely name it before another brand.

LaCroix runs the other way. It appears in 12 moments and is named first in 6. LaCroix scores 18 on recommendation and ranks fifth at 70. Liquid Death appears in 6 moments and is named first in 3. Both brands appear less often than Poppi and lead more often.

Spindrift and Olipop appear in 28 moments each, and each is named first in 10. A maximum recommendation score is a measure of presence, not a first-place count. Volume and primacy are separate assets. A brand can buy one of them and still miss the other.

KEY FINDING 05

Neither leading brand appears in the family favorite answers.

Spindrift leads the board and is absent from with breakfast, family favorite, cultural footprint and sourcing transparency. Olipop holds the maximum recommendation score and is absent from family favorite, real fruit, simple ingredients, natural and real, and for kids. Family favorite holds neither of the two leaders.

These questions are open lanes for the brands ARA measured. They are not empty answers. The models still return a name when a buyer asks for a family favorite or a drink for kids. The shelf shows where those names come from.

Bubly, Waterloo, Zevia, Vita Coco, Aura Bora, Culture Pop, Harmless Harvest and Health-Ade all hold choices on this shelf. ARA measured none of them. An open lane is held by someone the board does not list. A brand that wants that lane has to take it from a name already in the answer.

WHAT THIS MEANS

Thirteen of the twenty brands sit in SWELL. They hold equity the models under-reward. Milo's, Welch's, Natalie's, Uncle Matt's and Simply all score 59 or higher and take almost no recommendation. More clarity will not move those brands. The models are already clear about them. The work is to become the answer to a named question.

Three brands sit in EBB. Olipop, Poppi and LaCroix are recommended above the equity ARA measures. That position is an asset with a shelf life, because the models revise what they say. Spindrift, Liquid Death, Simple Truth and Naked sit where reward and equity agree. The median for this category is 56.5, and the board runs from 81 down to 47.