AI agents recommend Shopify products.
We get yours on the list.

SEO tools (Yoast, Ahrefs, SurferSEO) measure how Google's search crawler ranks pages. We measure what AI shopping agents (ChatGPT, Claude, Gemini, Mistral, DeepSeek - Perplexity joining Q3) actually recommend when our standardized buyer-intent query suite is issued daily. The two systems share some signals (structured data, alt text) but weight them very differently. SEO tools also can't tell you which products are missing from agent answers - we can. Note: our captures are a benchmark, not a sample of real shopper traffic (which is private to each agent).

Three structural differences: (1) we capture 5 AI agents synchronously on the same query (Perplexity coming Q3) - Profound watches 2-3, Athena AI watches ChatGPT only, ShopRank uses 4 dimensions vs our 8; (2) we prove fixes worked causally with DiD + Welch's t-test (p < 0.05) - they use observational attribution; (3) we offer a money-back Score Guarantee - none of them do. Pricing comparison: 40rty $49-499/mo fixed · ShopRank $9.99/100 SKUs · SKAW $19.99/mo · AI Catalog Score $49-399 with Performance tier at 5% of uplift (zero if no lift). Plus our scoring methodology is CC0 open source; theirs is closed.

No measurable impact on storefront speed. We use the Shopify Admin API server-side (off-thread, no shopper-facing JS). The only thing that touches your theme is an optional JSON-LD injection in product page headers - that adds ~2 KB to the HTML (smaller than a typical logo image). Tested against Core Web Vitals on 50+ stores: no LCP/FID/CLS regression.

Run the free 60-second scan first. If your score is already A+ (90 or above) across all 8 dimensions and 5 agents, you genuinely don't need us - we'll tell you that and suggest audit-only tier ($0). In practice, the average self-reported "optimized" catalog scores B- to C+ on first scan. Agentic commerce signals (JSON-LD product schema, metafield completeness, multi-agent reviews) are different enough from classic SEO that very few merchants are actually optimized for both.

Opposite effect. JSON-LD structured data, alt text, richer product metadata, and consistent metafields are universally good for search engines. Google indexes the same signals AI agents use - they just weight them differently. Beta merchants observed +3-8% in organic Google traffic in the 30 days following catalog optimization, alongside the AI lift.

We measure your AI-attributed revenue (sessions and orders coming from ChatGPT, Claude, Gemini, Mistral, DeepSeek - identified via referrer signals + User-Agent patterns + UTM matching) over a 30-day baseline before activation, then monthly after. The 5% only applies to the uplift over baseline, capped at $5,000/month. If sessions don't grow, your bill is $0. The causal model and attribution logic is documented in our open spec - you can verify our math.

AI agent attribution is harder than classic web analytics because referrers are inconsistent. We use a 3-signal stack: (1) ?utm_source=chatgpt-style explicit UTMs when agents pass them; (2) User-Agent matching for known agent crawlers (OAI-SearchBot, PerplexityBot, ClaudeBot, Google-Extended, Mistral-AI, DeepSeek-Crawler); (3) temporal correlation with our ground-truth captures - when our pipeline records a recommendation of your product at time T and your store gets a session referrer-less or with a generic referrer within ±10 minutes, we attribute it.

Honest about limits: we capture roughly 60-80% of AI-attributed traffic. Some agent outputs don't leave a trail at all (voice, screenshot, copy-paste). We err conservative, so our 5% performance fee is calculated on what we can prove - never inflated. The full methodology is in the CC0 spec.

Fair question. Yes, you can read our CC0 spec, fork our MIT skills, and self-audit your catalog manually. That's the point - we don't want to hostage you. What you pay for is tooling, automation, and proof:

(1) Multi-agent synchronous capture - querying all 5 agents at the same moment, 2,400 times a day, on the same query. Trivial to describe, ~80-120h of infra to build (rate-limits, retries, billing, agent API differences). (2) The dataset itself - 79k+ captures grow by 2,400/day, you can read it but you can't recreate it without our pipeline running for months. (3) The causal A/B framework - DiD + Welch's t-test on session deltas, control split, p-value reporting. Doable but ~40h of stats work. (4) Apply-into-Shopify - reading the spec is easy, but writing automated fixes with dry_run guardrails, audit journaling, 30-day undo, GDPR compliance is real product work. (5) Score Guarantee - we put +10 pts/30 days in the ToS. No one self-auditing has a refund mechanism on themselves.

Plenty of merchants will read the spec and self-fix. We're betting most prefer to pay the tooling fee and skip the build.

The free plan lets you audit up to 50 SKUs with the full scoring engine, including the AI Readiness Score, Agentic Checklist, and AI Query Simulator (3 queries/day). No credit card required, no expiry, no feature throttling on the 50 SKUs you choose. When you outgrow it, upgrade to Growth ($49/mo, 500 SKUs), Pro ($149/mo, 3,000 SKUs), or Performance ($399/mo, unlimited SKUs) directly from the Shopify admin.

Today the published price is the price: $49 / $149 / $399 per month, or 30% off when billed yearly. We don't run hidden-trial games, surprise upsells, or a price-creep schedule baked into the terms. If we ever change list pricing, existing subscribers stay on their current rate for the duration of any active billing period and get clear notice before any renewal at a new price.

If your AI Catalog Score doesn't lift by at least +10 points within 30 days of activating a paid plan (and you applied at least the top 5 recommended fixes), every dollar you paid in that window is refunded. No support ticket required - we auto-refund when the threshold isn't met. The guarantee is in the Terms of Service v1.0, not the marketing copy.

No. Every mutation runs with dry_run: true by default. We preview the exact change, you click approve, then we apply. Every action is journaled per-SKU with a 30-day full undo. We never overwrite product titles or descriptions without explicit consent, never touch pricing or inventory, and never modify any field you've marked as locked.

Causal A/B testing using Difference-in-Differences with Welch's t-test on per-product session deltas. We split your products 50/50 treatment/control, measure AI-attributed sessions before and after the fix, and only declare uplift when p < 0.05. Every other tool gives you observational attribution ("correlation"). We give you statistical proof ("causation"). Full methodology in the CC0 spec.

Read access: products (titles, descriptions, images, metafields, variants), order metadata (for AI traffic attribution - no customer PII), and theme files (for optional JSON-LD injection). Write access: only the fields you explicitly approve in the dry-run preview. We never store customer names, emails, phone numbers, or addresses. Full privacy policy at privacy.html.

Yes - the 8-dimension scoring spec is published under CC0 1.0 (public domain) at commerce-agentic/agentic-commerce-spec. The reference TypeScript implementation, scoring weights, and ML uplift models are MIT. You can fork it, audit it, build a competing product, or wait for Shopify/BigCommerce/WooCommerce to adopt it in their official docs. We make money on the tooling, automation, and proof - not the score itself.

Yes. On uninstall, all data is purged from our database within 48 hours per Shopify's GDPR webhook requirements (we honor app/uninstalled, shop/redact, and customers/redact). Score metafields written to your products stay there (you own them on your Shopify), but you can delete them manually via the admin or with one API call - we provide the script in our docs.