ARA — Methodology

What ARA Measures

Most AI-and-brands work today is solving for either content (Wave 1 — faster ads) or discovery (Wave 2 — be findable). Real and necessary; both downstream. ARA measures Wave 3 — coordination: the AI agents now standing between demand and supply, interpreting what customers want, framing the choice set, and executing the recommendation.

In that world the strategic question isn't whether AI can find your brand. It's whether AI carries the right frame for it — built from the category meaning you intended, not the category default that gets compressed into the gap. ARA is the audit of how that frame has been deposited into the systems people now use to decide what to buy.

Five dimensions, scored 0–20 each. Four frontier models — Claude Sonnet, GPT‑4o, Perplexity sonar‑pro, Gemini 2.5‑flash — plus a live agent‑readiness scan against your owned web infrastructure. Real responses, real prompts, real evidence per cell.

Tier Definitions

83–100

Awesome

Category authority. AI carries the brand's intended frame and chooses it across the demand moments that matter most.

70–82

Strong

Reliably chosen by AI in primary recommendation contexts. Frame is mostly intact but not yet category-defining.

56–69

Average

Surfaced when prompted, but AI defaults to category-generic framing. Brand is named, not advocated for.

40–55

Weak

AI carries someone else's frame for the brand — competitor framing, owner persona, or stale narrative dominates.

0–39

Invisible

Does not surface in relevant demand moments. No distinct frame in machine memory.

The Five Dimensions

Code	Dimension	Max Score	What It Measures	Key Signals
A1 · STR	Structural Readiness	20 pts	Whether AI can retrieve you at all. The structural floor: official sources, machine-parseable data, schema markup, and the agent-readiness infrastructure (robots.txt, llms.txt, MCP, A2A) that determines whether AI systems can read and act on your brand at all.	Schema.org markup depth First-party data infrastructure E-commerce and app data continuity Structured product and service data
A2 · SEM	Semantic Clarity	20 pts	Whether AI tells your story — or someone else's. The semantic question: does AI describe your brand in your own vocabulary and positioning, or does it default to category-generic language, competitor framing, or owner persona?	Owned vocabulary and positioning language Ingredient / attribute specificity Consistency across digital touchpoints Occasion and use-case clarity
A3 · SYN	Synthetic Customer Test	20 pts	Whether AI chooses you in the moments that matter. Reciprocal-rank scoring across category-level demand queries (no brand mentioned in the prompt) on four frontier models. The single best evidence of whether your frame is winning.	AI recommendation win rate by query type Recommendation confidence score Breadth of occasions won Consistency across AI platforms
A4 · EMO	Emotional Residue	20 pts	Whether AI carries your cultural narrative, or a competitor's. Brands with rich communities, partner ecosystems, and earned cultural moments have been continuously training AI on their own behalf — without ever intending to. A4 measures how much of that signal landed.	Community-generated content volume and quality Sentiment intensity and consistency Cultural association depth Brand advocacy patterns in training data
A5 · VOI	Archetype, Personality & Voice	20 pts	Whether your voice survives when the screen disappears. Voice assistants, chat agents, and agentic commerce strip away every visual asset. A5 measures how recognisable, distinct, and attributable your brand remains in pure language — and whether AI can speak it back without losing the register.	Voice distinctiveness and consistency Phonetic clarity of brand and product names Tone consistency across indexed content Voice assistant surfacing quality

A1 · Structural Readiness Rubric

Whether AI can retrieve you at all. / 20

17–20

Awesome

Proprietary ingredient data, SKUs, revenue figures, and certifications are machine-parseable. Structured data schema is comprehensive and current. Brand generates rich first-party data continuously.

14–16

Strong

Brand category, parent company, founding year, and key campaigns are correctly structured and indexed. Schema markup covers core entities. Data infrastructure is present and mostly coherent.

11–13

Average

Brand exists in structured data and is correctly categorised, but proprietary or differentiating attributes are absent. Machines can find the brand but lack the specifics needed for confident recommendation.

8–10

Weak

Structured data conflates the brand with competitors or related entities. Machine parsing produces ambiguous or inaccurate results. Category assignment may be incorrect or contested.

0–7

Invisible

Machine systems cannot differentiate the brand from its category. No distinct structured data presence. Brand is effectively invisible to recommendation infrastructure.

A2 · Semantic Clarity Rubric

Whether AI tells your story — or someone else's. / 20

17–20

Awesome

Machine systems use the brand's own vocabulary, archetype, and positioning language when describing it unprompted. The brand owns a distinct semantic territory that competitors do not encroach on.

14–16

Strong

Machine description is accurate and clearly positions the brand distinct from competitors. Owned vocabulary is reflected in AI outputs. Positioning is legible and consistently applied across platforms.

11–13

Average

Machine description is relational — the brand is defined in comparison to a competitor rather than on its own terms. Positioning exists but is not owned. Semantic territory is contested.

8–10

Weak

Machine systems define the brand primarily as "an alternative to X." The brand has no independent semantic identity — it exists only in relation to a stronger competitor in machine memory.

0–7

Invisible

Machine description contradicts brand intent, or the brand is absent from relevant semantic associations. AI systems attribute incorrect positioning or personality to the brand.

A3 · Synthetic Customer Test Rubric

Whether AI chooses you in the moments that matter. / 20

17–20

Awesome

Brand is surfaced unprompted across 5 or more query types. Consistently recommended in top-1 or top-2 position. Strong social signal boosts recommendation confidence. Wins across multiple purchase occasions.

14–16

Strong

Recommended across 3–4 query types. Wins at least one distinct purchase occasion clearly. Recommendation is consistent across AI platforms but not dominant across all contexts.

11–13

Average

Recommended in 1–2 query types. Appears in secondary tier only — listed but not prioritised. Machines know the brand exists but do not lead with it in competitive contexts.

8–10

Weak

Brand only surfaces when its category is named directly in the query. Not recommended in contextual or occasion-based queries. AI agents do not associate the brand with specific use cases.

0–7

Invisible

Brand is absent from all synthetic query testing. AI agents consistently recommend competitors in every tested occasion and context. Effectively zero AI-driven recommendation presence.

A4 · Emotional Residue Rubric

Whether AI carries your cultural narrative, or a competitor's. / 20

17–20

Awesome

Brand owns a specific, named emotional territory. Multiple cultural moments are cited by machine systems. The brand has transcended its category — AI associates it with a feeling, community, or life context, not just a product.

14–16

Strong

Clear emotional associations are present and consistent. At least one cultural moment or community touchpoint is cited unprompted by machine systems. Sentiment is positive and differentiated from generic category affect.

11–13

Average

Machine systems associate the brand with generic emotions such as "fun" or "youthful" but cannot name specific moments, communities, or cultural contexts. Emotional associations exist but are not ownable or distinct.

8–10

Weak

Emotional description is entirely relational — machines describe the brand as "feels different from [competitor]" with no independent emotional identity. Brand's emotional residue exists only in contrast, not in its own right.

0–7

Invisible

No emotional associations present. Machine systems describe the brand in purely functional terms. No evidence of brand-building investment in machine-indexed content. Pure utility, zero affect.

A5 · Archetype, Personality & Voice Rubric

Whether your voice survives when the screen disappears. / 20

17–20

Awesome

Brand identity is entirely verbal and tonal. A specific archetype is named by machine systems unprompted. The brand's personality survives complete visual removal — it remains distinct and attributable in pure text or audio contexts.

14–16

Strong

Brand personality is a mix of verbal and visual assets, but at least one strong verbal or tonal asset is identified by machines. Archetype is discernible and attributable without visual cues in most contexts.

11–13

Average

Brand archetype is discernible but relies heavily on references to visual platforms or assets. Machine description of personality includes frequent references to what the brand looks like rather than how it sounds or speaks.

8–10

Weak

No clear archetype is present. Brand personality collapses entirely without visual references. Machine systems cannot describe a consistent voice or tone — personality is platform-dependent and non-transferable.

0–7

Invisible

No audio or verbal personality is detectable. Brand is interchangeable with category competitors in any non-visual medium. Machine systems apply generic tonal descriptors with no brand-specific attribution.

Audit Process

01

Category Scoping

Define the competitive set, relevant query universe, and geographic scope. Identify the key purchase occasions and recommendation contexts that matter most for the category.

02

Structural Audit

Technical review of schema markup, data infrastructure, first-party data generation, and digital channel coherence across brand properties. Benchmarked against category peers.

03

Synthetic Query Testing

Systematic prompting of leading AI platforms (ChatGPT, Claude, Gemini, Perplexity) across the query universe. Win rates, confidence levels, and recommendation rationale are captured and scored.

04

Scoring and Recommendations

Dimension scores are assigned and combined into the overall ARA score. Findings, vulnerabilities, and a prioritised recommendation roadmap are developed for each brand in the study.

Important Note on Scoring

ARA Index scores reflect what AI is choosing about your brand at a specific point in time. Training data evolves continuously — brands that deposit a clear frame and rich interaction signal compound their score, while brands without it get compressed into the category default a little further with every cycle. The window to act is now, before the next training cycle locks in the current state.

The ARA Index is a proprietary methodology developed by araco.ai. Scoring combines structural analysis, semantic evaluation, reciprocal-rank synthetic query testing across four frontier models, emotional-residue assessment, voice-survival analysis, and a live agent-readiness scan. All brand audits are confidential and intended for the commissioning client only.