Methodology
The Agentic Readiness Index scores Shopify brands on how prepared their site, product data, and external footprint are to be discovered, evaluated, and recommended by AI shopping agents — ChatGPT, Gemini, Perplexity, Claude. The rubric is open and the scoring is deterministic.
What we score
Six dimensions, each a deterministic check against the brand's live site or its external citation footprint. No LLM guesses, no opinion-based scores.
Schema.org Product Markup — 20% weight
AI shopping agents build a product card from structured data. If your Product schema is missing required fields, you either don't appear in the agent's results or appear with incomplete information. We check for Product schema, required and recommended fields (name, image, description, offers, GTIN/MPN, aggregateRating, Review, FAQ, variants) on a representative product page.
AI Crawlability — 15% weight
If your robots.txt or WAF blocks AI bots, you're invisible to them regardless of how clean your data is. We check robots.txt for explicit AI bot rules (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, Applebot-Extended), test whether each bot's user agent receives a successful response when fetching your homepage, and look for an llms.txt at your root.
PDP Context Depth — 15% weight
Agents need substantive context to compare and recommend products. We measure word count of meaningful product copy (excluding nav and footer), presence of structured spec tables, visible FAQ sections, review schema with body text, and alt-text density on product images.
Commerce Protocol Readiness — 10% weight
Direct in-chat purchase is rolling out through OpenAI ACP, Google UCP, and WebMCP — all built on top of major commerce platforms. We detect your underlying platform (Shopify, BigCommerce, WooCommerce), check for Stripe integration (required for ACP), verify your products surface in Google Shopping for category searches, and cross-check against published protocol partner lists.
Third-Party Trust Signals — 20% weight
AI engines cite third-party authorities heavily — roughly 90-95% of their reference sources are off-site. We measure your presence on Trustpilot (review count + rating), Reddit (discussion volume in last 12 months), major publications (NYT, Wirecutter, Strategist, Vox, The Verge, WSJ), and recent press coverage.
AI Surface Test — 20% weight
The outcome layer. For category pages we run five shared shopping prompts per category across five web-grounded models (OpenAI, Anthropic, Gemini, Perplexity, Grok). For individual brand reports we derive two natural shopping queries that match the product category (without naming the brand). We measure how many trials surface the brand, and whether it appears early in the response (proxy for top recommendation) or late.
AI demand estimation (category pages)
Category pages estimate AI shopping prompt demand from historical Google query volumes, then use that estimate to weight Share of Voice. For each test prompt we map to the closest US keyword and fetch average monthly search volume via Google Ads Keyword Planner.
We apply an AI-to-Google adoption ratio of 10% — the midpoint of an 8–12% range observed for shopping-style queries as of May 2026. Estimated AI prompt volume = Google monthly volume × 10%.
Important: Google Keyword Planner volume buckets are rounded (typically to values like 50, 500, 5,000, 50,000). Treat category and prompt demand estimates as order-of-magnitude directional signals, not exact counts.
Weighted Share of Voice = ∑(mention rate in prompt i × estimated AI prompt volume of prompt i) / ∑(estimated AI prompt volume). Raw mention rate (unweighted across prompts) is shown alongside for transparency. Keyword mappings are human-reviewed; volumes are cached ~30 days.
Brand awareness measurement
Category pages can include a Brand awareness vs AI visibility chart. Brand search volume is the sum of US Google Keyword Planner monthly volumes for a curated set of brand-name queries per domain (e.g. “allbirds”, “allbirds shoes”). Keywords with no measurable volume in Keyword Planner are omitted; brands with zero matched volume are labeled below KP measurement (<50 searches/mo).
AI Fairness ratio compares weighted AI Share of Voice to log-normalized brand search fame within the category:
awareness_normalized = log(brand_search_volume + 1) / log(max_brand_search_in_category + 1)
AI Fairness = weighted_SoV% / max(awareness_normalized × 100, 1)
Buckets: BONUS (≥1.3× expected), ALIGNED (0.5–1.3×), PENALTY (<0.5×). Brands below measurement threshold are not bucketed.
Data quality: Google Keyword Planner rounds volumes to coarse buckets (50, 500, 5,000, 50,000, etc.) — treat all brand and category demand figures as order-of-magnitude signals. Saola is flagged suspect because the brand name overlaps with a critically endangered antelope species; its volume may include non-brand intent and is excluded from auto-generated strategic callouts.
Category keyword data — sources and full keyword list
Source: Google Keyword Planner exports (US, monthly averages). The table below shows the top 50 keywords by monthly volume per category. Est. AI prompts/mo uses the same 10% AI-adoption ratio used throughout category pages.
Sustainable Footwear
| Query | Est. AI prompts/mo | AI demand trend | |
|---|---|---|---|
| 1 | sustainable footwear | ~5,000/mo | +9,900% |
| 2 | sustainable shoes | ~5,000/mo | +900% |
| 3 | sustainable sneakers | ~5,000/mo | — |
| 4 | sustainable tennis shoes | ~5,000/mo | — |
| 5 | dr martens vegan | ~500/mo | — |
| 6 | durable sneakers | ~500/mo | +900% |
| 7 | earth friendly shoes | ~500/mo | +900% |
| 8 | eco friendly footwear | ~500/mo | — |
| 9 | eco friendly shoes | ~500/mo | +900% |
| 10 | eco friendly sneakers | ~500/mo | +9,900% |
| 11 | eco friendly tennis shoes | ~500/mo | +9,900% |
| 12 | eco friendly trainers | ~500/mo | +9,900% |
| 13 | environmentally conscious shoes | ~500/mo | +900% |
| 14 | environmentally friendly footwear | ~500/mo | — |
| 15 | environmentally friendly shoes | ~500/mo | +900% |
| 16 | environmentally friendly slippers | ~500/mo | +900% |
| 17 | environmentally friendly sneakers | ~500/mo | +9,900% |
| 18 | environmentally friendly tennis shoes | ~500/mo | +9,900% |
| 19 | ethical athletic shoes | ~500/mo | — |
| 20 | ethical gym shoes | ~500/mo | — |
| 21 | ethical sneakers | ~500/mo | — |
| 22 | ethical tennis shoes | ~500/mo | — |
| 23 | friendly shoe | ~500/mo | — |
| 24 | ladies vegan shoes | ~500/mo | — |
| 25 | margielas sneakers | ~500/mo | — |
| 26 | nike's move to zero | ~500/mo | -90% |
| 27 | sneakers ethical | ~500/mo | — |
| 28 | sustainable footwear brands | ~500/mo | +900% |
| 29 | sustainable slip on shoes | ~500/mo | +99,900% |
| 30 | sustainable womens sneakers | ~500/mo | — |
| 31 | vegan doc martens | ~500/mo | — |
| 32 | vegan leather sneakers | ~500/mo | — |
| 33 | vegan sambas | ~500/mo | -90% |
| 34 | vegan shoes women | ~500/mo | — |
| 35 | vegan sneakers | ~500/mo | — |
| 36 | vegetarian doc martens | ~500/mo | — |
| 37 | vegetarian dr martens | ~500/mo | — |
| 38 | vegetarian shoes | ~500/mo | — |
| 39 | vegetarian sneakers | ~500/mo | — |
| 40 | will's vegan shoes | ~500/mo | — |
| 41 | women's vegan sneakers | ~500/mo | — |
| 42 | 1460 doc martens vegan | ~50/mo | — |
| 43 | 1460 dr martens vegan | ~50/mo | — |
| 44 | adidas allbird | ~50/mo | — |
| 45 | adidas samba vegan | ~50/mo | — |
| 46 | adidas samba vegan shoes | ~50/mo | -90% |
| 47 | adidas superstar vegan | ~50/mo | — |
| 48 | adidas vegan shoes | ~50/mo | — |
| 49 | adidas vegan sneakers | ~50/mo | — |
| 50 | adidas vegan trainers | ~50/mo | — |
Raw mention rates
Full per-brand mention metrics for each category prompt run. Weighted % uses prompt-demand weighting; Raw % is the unweighted mention rate across the shared prompt matrix.
Non Alcoholic Cocktails
| Domain | Weighted % | Raw % | Prompt breakdown |
|---|---|---|---|
| lyres.com | 80% | 80% | No prompt data |
| seedlipdrinks.com | 80% | 80% | No prompt data |
| curiouselixirs.com | 70% | 70% | No prompt data |
| drinkghia.com | 60% | 60% | No prompt data |
| kineuphorics.com | 50% | 50% | No prompt data |
| athleticbrewing.com | 40% | 40% | No prompt data |
| aplos.world | 0% | 0% | No prompt data |
| drinkhiyo.com | 0% | 0% | No prompt data |
| wearedaytrip.com | 0% | 0% | No prompt data |
| takearecess.com | 0% | 0% | No prompt data |
Prebiotic Soda
| Domain | Weighted % | Raw % | Prompt breakdown |
|---|---|---|---|
| drinkolipop.com | 100% | 100% | No prompt data |
| drinkpoppi.com | 80% | 80% | No prompt data |
| drinkwildwonder.com | 40% | 40% | No prompt data |
| slicesoda.com | 10% | 10% | No prompt data |
| drinkpopwell.com | 10% | 10% | No prompt data |
| drinkdroplet.com | 0% | 0% | No prompt data |
Sustainable Footwear
| Domain | Weighted % | Raw % | Prompt breakdown |
|---|---|---|---|
| allbirds.com | 98% | 92% | P1: 100%, P2: 100%, P3: 80%, P4: 100%, P5: 80% |
| veja-store.com | 91% | 96% | P1: 80%, P2: 100%, P3: 100%, P4: 100%, P5: 100% |
| rothys.com | 83% | 60% | P1: 100%, P2: 80%, P3: 0%, P4: 80%, P5: 40% |
| nativeshoes.com | 63% | 76% | P1: 80%, P2: 40%, P3: 80%, P4: 80%, P5: 100% |
| on.com | 46% | 32% | P1: 60%, P2: 40%, P3: 20%, P4: 20%, P5: 20% |
| thousandfell.com | 38% | 32% | P1: 40%, P2: 40%, P3: 20%, P4: 60%, P5: 0% |
| nisolo.com | 27% | 16% | P1: 60%, P2: 0%, P3: 0%, P4: 20%, P5: 0% |
| wills-vegan-shoes.com | 23% | 32% | P1: 40%, P2: 0%, P3: 40%, P4: 0%, P5: 80% |
| saola.com | 18% | 12% | P1: 20%, P2: 20%, P3: 0%, P4: 20%, P5: 0% |
| indosole.com | 18% | 12% | P1: 40%, P2: 0%, P3: 0%, P4: 0%, P5: 20% |
| nothingnew.com | 10% | 12% | P1: 0%, P2: 20%, P3: 0%, P4: 20%, P5: 20% |
| etiko.com.au | 9% | 4% | P1: 0%, P2: 20%, P3: 0%, P4: 0%, P5: 0% |
| hyloathletics.com | 9% | 4% | P1: 20%, P2: 0%, P3: 0%, P4: 0%, P5: 0% |
| 8000kicks.com | 1% | 4% | P1: 0%, P2: 0%, P3: 0%, P4: 0%, P5: 20% |
| tropicfeel.com | 0% | 0% | P1: 0%, P2: 0%, P3: 0%, P4: 0%, P5: 0% |
| feelgrounds.com | 0% | 0% | P1: 0%, P2: 0%, P3: 0%, P4: 0%, P5: 0% |
How we score
Every check is deterministic. Each dimension has explicit numeric thresholds tied to underlying data — if we find a Product schema with all required fields, your score for that dimension is high. If we don't, it isn't. We do not let language models invent check results — they only write the explanation prose around results that came from parsing actual data or counting API responses.
This matters because if you push back on a score, we can show you exactly what we found and what we didn't. No hallucinated reasons.
What we don't do
- We do not crawl your site repeatedly. Each score is a single point-in-time snapshot.
- We do not analyze adult or restricted product categories. AI engines filter these regardless of agent-readiness.
- We do not provide advisory or implementation services. This index is a measurement tool, not a consultancy.
Want your brand re-scored?
If you've made changes since we last looked, or want to be added to the index, email hello@agenticreadyshop.com.