FAQs.

Three disciplines, three levels of the stack.

SEO optimises for search crawlers — links, keywords, technical signals. Can machines find your pages?

GEO (Generative Engine Optimisation) optimises content for AI retrieval — structured articles, FAQs, product descriptions. Can machines retrieve your content? JBL's 2,434% LLM-referral surge during peak 2025 came from this layer.

ARA works upstream of both. We audit the brand itself — architecture, positioning, semantic identity, emotional footprint, voice. Do machines understand, trust, and recommend your brand? A brand with weak architecture hits a ceiling on GEO no matter how well its content is structured. ARA finds and fixes that ceiling.

AI is now the first point of discovery for a meaningful share of purchase decisions. When someone asks a language model what to buy or which brand to trust, they get an answer. That answer isn't random — it reflects what AI has absorbed from public data.

Brands that have invested in machine legibility appear in those answers. Brands that haven't are losing recommendations they don't know they're missing.

The stakes rise as AI agents gain the ability to execute purchases autonomously. The agent picks; the agent buys; no human at checkout. The brands machines don't know, trust, and recommend simply won't be bought.

Four pillars decide whether a brand rides The Swell. Five instrumented dimensions measure them.

Every score traces back through a chained audit trail. For a 10-brand study: When a CMO asks "why does my brand score 14 on emotional residue?", we show the four AI responses, the specific phrases, the rubric anchor those phrases mapped to, and the second independent run that confirmed the call.

No "trust me" scores. Every claim defended.

A standardised protocol holds every variable constant.

Query sets are designed before testing begins and held constant across all brands and platforms. Each AI conversation runs in a fresh session, no system prompt, default model settings, verbatim response recording. Brand mentions are coded against an explicit rubric, not analyst intuition.

Methodology is versioned and dated. Every study is locked to a specific methodology snapshot — old studies remain fully reproducible. Methodology changes are themselves logged: who changed what, when, why.

We validate across two frontier model generations (Claude Sonnet 4.6 and Opus 4.7). Brand rankings match exactly; 95% of sub-dimension scores within ±2 points of each other.

The Swell is where the energy of a category will meet consumer demand in the next 12 to 18 months. AI is shaping that future now — absorbing brand signals, forming the category models that will shape what consumers see when they ask… deciding who wins and who falls off a cliff.

Three forces are converging inside The Swell.

The data shaping AI outputs is being written now. AI systems don't discover brands in real time. They synthesise understanding from the cumulative record of what's been published, reviewed, discussed, and linked. A brand depositing machine-legible signal today will see it enter AI training cycles over the next several quarters. A brand that waits will be catching up to competitors who didn't.

Agentic AI is moving from experiment to product. MCP servers, A2A agent cards, .well-known endpoints — the technical layer that lets AI agents discover, compare, and transact on behalf of consumers is shipping now. In our Beverages study, Shopify-native DTC brands outscored Coca-Cola and Pepsi on agent-readiness without dedicated engineering effort. That gap will widen.

The window will close. AI training cycles run quarterly to annually. The brands ready when AI becomes the default discovery layer are the brands who measured and built during The Swell.

Yes. At two layers.

Reading The Swell. A category audit measures every major brand in a competitive set at the same moment. The patterns across the cohort — where AI is converging on certain mechanisms or claims, where consumer-side signal is building, where agentic infrastructure is moving — show you the energy forming offshore. A single-brand audit is a snapshot. A category audit is the weather report.

Diagnosing whether your brand will catch it. That's the per-brand output. Across Visibility, Comprehension, Fidelity, and Recommendability, ARA tells you where you stand today, where the gaps are, and what to deposit, fix, or codify over the next 12 to 18 months so AI absorbs you while The Swell is still building.

The public index at araco.ai covers broad competitive sets across published categories. Every brand in a published category has a live score, tier, and per-dimension breakdown. Independent ARA research, published as market intelligence.

A commissioned study is private by default and adds:

· Extended testing — more purchase occasions, query variants, platform-specific analysis
· Full per-brand findings: Strengths, Vulnerabilities, Opportunities, Watches
· Prioritised recommendations roadmap with implementation timeframes
· Category-level intelligence on AI positioning dynamics
· Wave-on-wave tracking for ongoing engagements

The public score is real. The commissioned study is everything you need to act on it.

A standard category study runs three to four weeks from scoping to delivery — query design, platform testing, structural and semantic analysis across all brands, and findings synthesis.

Pricing is scoped per engagement, based on category complexity, the size of the competitive set, and whether the study includes implementation support or ongoing wave tracking. A standard commissioned study is priced comparably to a mid-range research project — meaningful, but well within a marketing or strategy budget.

The public index is free.

If your category is already in the public index, your baseline is live at araco.ai. When you're ready to go deeper, request an audit. We'll be in touch within two business days to scope the commissioned study.

We don't need access to your internal systems or data. Our methodology is entirely external — we assess how AI systems represent your brand based on public information and live query behaviour.