The AICE Score answers one simple question: will AI shopping agents find, understand, and trust your products enough to recommend them? This article explains the signals we measure and where small merchants can get the most impact quickly.
The six pillars
Our diagnostic groups 54 signals into six pillars. Each pillar maps to a step in the recommendation pipeline — discovery, understanding, trust, conversational fit, transaction readiness, and feed health.
1) Discoverability
Can the crawler reach your content? Checks include robots.txt, llms.txt presence, sitemap completeness, canonical URLs, and page speed for core product pages. If crawlers are blocked, nothing else matters.
2) Product Intelligence
Do product pages include structured data (Schema.org JSON-LD), clear attribute fields (brand, material, size), and consistent titles? AI needs machine-readable facts to compare alternatives.
3) Trust Signals
AggregateRating schema, merchant contact and identity in structured form, explicit returns and shipping policies — these help AI filter for reputable sellers.
4) Conversational Readiness
Do your pages answer the kinds of questions people ask an assistant? We score for FAQ schema, clear "best for" language, and attribute-based comparisons.
5) Transaction Frictionlessness
Can an AI-referred buyer actually complete a purchase? We check for guest checkout, clear shipping and returns, and consistent availability data in product feeds.
6) Feed & Integration Health
Is your product data flowing to the catalogue endpoints agents read? We verify feed freshness, unique identifiers (GTIN/SKU), and platform integration flags like Shopify Agentic storefronts.
Based on our own audit work, we've observed that stores typically gain meaningful AICE points by addressing three common gaps: adding JSON-LD product schema, creating or verifying llms.txt availability, and making reviews machine-readable with AggregateRating.
How scoring works
Each of the 54 signals is weighted by impact. Critical signals (crawler access, schema presence) have higher weight. Scores are normalised to a 0–100 scale and bucketed into Green/Amber/Red zones.
Quick implementation checklist
Want a free AICE audit?
Send the store URL to jay@aicescore.com and receive a short report with pillar breakdown and prioritised fixes.