AI Recommendation Patterns: Performance Beverage

Ally Kiel · AI Discoverability Strategy for Premium F&B Brands · March 2026

Key Findings

• Recess and Kin dominate all clusters — 31 and 29 appearances across 58 data points

• Hiyo is clear #3 at 19; Ghia and De Soi tied at 16, each for different reasons

• Ghia's appearances are driven by occasion-based authority, not functional benefit association

• Non-Alcoholic Canned surfaces a distinct competitive set from all other clusters

• Energy/Focus is the most fragmented cluster — the largest open opportunity in the category

Context

The Recommendation Layer

AI discovery tools are increasingly answering product questions directly instead of returning search results. When consumers ask questions like "best adaptogenic drinks" or "healthy energy drink alternatives," AI tools return condensed recommendation sets — typically 3 to 6 brands — rather than lists of websites. This creates a new competitive dynamic: brands must not only exist in the category — they must be included in AI-generated answers. For performance beverage founders, the question is no longer only "can consumers find us?" but "does AI recommend us?"

How Recommendation Clusters Form

Large language models generate responses by synthesizing patterns learned from massive datasets — brand mentions across articles, buying guides, editorial recommendations, product comparisons, and structured product data. Over time, models learn that certain brands frequently appear in similar contexts, producing stable semantic brand clusters. Once a cluster forms, it tends to reinforce itself. Brands outside the cluster may appear inconsistently or not at all — a phenomenon that can be described as recommendation gravity.

Research Scope

Platforms tested: ChatGPT, Google Gemini

Total prompts: 29 prompts across 5 intent clusters

Total data points: 58 (29 prompts × 2 platforms)

Prompt approach: Purchase-intent and benefit-driven queries that generate brand recommendations

Clusters tested: Social/NA · Stress/Calm · Energy/Focus · General Performance Beverage · Non-Alcoholic Canned

Overall Findings

Default Brand Formation

Across 58 data points, a small group of brands appears with striking consistency. The drop-off after the top two is steep — confirming that AI recommendation clusters are already consolidating around a small set of defaults.

Cluster Analysis

How the Category Splits by Intent

Performance beverage prompts do not return a single consistent brand set. AI systems respond differently depending on the consumer's intent. Five clusters reveal meaningfully different competitive landscapes.

Social / Non-Alcoholic — Most Consolidated

Recess and Kin each appear across all 5 social prompts on both platforms — a perfect 10/10. Ghia is #3 with 6 appearances, yet carries no functional benefit association. Its authority is purely occasion-based: AI systems have absorbed years of food media, lifestyle editorial, and sommelier culture positioning Ghia as what a certain kind of person drinks instead of alcohol. This is a significant finding — occasion-based positioning can be as powerful as benefit-based positioning for social intent prompts.

Stress / Calm — Most Contested

The most active competition below the top two. Trip, Melting Forest, Hiyo, and Moment all have meaningful presence. Hiyo's 5 appearances — despite being positioned primarily as a social brand — suggests AI is classifying it across multiple benefit territories. De Soi also appears here, reinforcing its unusual cross-cluster consistency.

Energy / Focus — Most Fragmented

The most strategically significant cluster for emerging performance brands. Functional adaptogenic beverages compete directly with legacy energy brands — Celsius, Alani Nu, Bloom — and AI systems have not yet formed stable associations favoring the premium functional set. Heywell's appearances are format-driven (sparkling + adaptogens) rather than benefit-driven. Odyssey Elixir appears on both platforms — a brand gaining cross-platform traction. This cluster represents the largest open opportunity in the category.

General Performance Beverage — Broadest Spread

General wellness and adaptogen prompts return the widest brand spread. Four Sigmatic appears 5 times — indicating strong mushroom/adaptogen authority that doesn't transfer to the canned functional space. Format matters: category authority in one form doesn't automatically carry to adjacent formats. BREZ appears 4 times here. Worth noting: BREZ's product line spans both adaptogenic and THC-infused beverages — its AI visibility likely reflects its adaptogenic and mushroom-based SKUs, which AI systems can recommend without restriction.

Non-Alcoholic Canned — A Distinct Competitive Set

This cluster surfaces a notably different competitive landscape from the other four. Ghia dominates with 6 appearances — its highest concentration in any cluster — driven entirely by its cocktail culture and food media presence rather than functional benefits. St. Agrestis appears consistently here and essentially nowhere else in the dataset. This confirms that non-alcoholic canned brands compete on a different discovery pathway from functional performance beverages — one rooted in drinking occasion and cultural identity.

Platform Divergence

ChatGPT vs. Gemini

ChatGPT and Gemini do not return the same brand sets — sometimes significantly. Visibility on one platform does not guarantee visibility on the other, which has direct implications for where brands invest in signal building.

A brand appearing consistently on ChatGPT but not Gemini — or vice versa — has a diagnosable signal gap. The sources each platform draws from differ, and benefit positioning needs to be reinforced across enough surfaces to achieve cross-platform visibility.

Brand Positioning

Where does your Brand Stand?

Based on this research, performance beverage brands currently fall into one of three positions.

These positions are not permanent. Recommendation clusters are still forming in this category. The brands that move from inconsistent to default over the next 12 to 18 months will likely be those that intentionally align their benefit positioning with how AI systems learn to classify them — and that build signal consistently across both platforms, not just one.

The Window Is Open — For Now

The performance beverage category is in an early and consequential window.

AI recommendation clusters are not yet fully calcified. A small number of brands have established default positions — but the majority of the category remains in flux. That creates a genuine opportunity for brands that understand how these systems form associations and act before the window closes. The findings in this report point to a clear pattern: benefit clarity and signal consistency drive recommendation inclusion more reliably than brand recognition or distribution reach. A brand with strong retail presence but diffuse messaging may be invisible in AI-generated discovery. A smaller brand with sharp, consistent benefit positioning may outperform it. What This Means Practically Founders should be asking: when a consumer asks an AI assistant for a recommendation in our category, does our brand appear? If not — or only sometimes — the gap is diagnosable and addressable. The signals AI systems learn from are not abstract. They live in product descriptions, editorial coverage, benefit-forward content, and retail metadata. Brands that audit their current signal footprint and align it with how these systems classify the category will be better positioned as AI-assisted discovery continues to grow.