Skip to content

Step 1 — Diagnosis

The diagnosis is a scored audit of your entire product catalog. It runs in under four minutes for stores with up to 250 products and produces a detailed, actionable report.

What gets checked

Findability baseline

Before looking at individual products, Agentic Prep checks whether AI crawlers can even reach your store:

  • robots.txt — are AI crawlers blocked?
  • sitemap.xml — is there a valid sitemap for product pages?
  • Homepage canonical/noindex — are there signals that could prevent indexing?
  • llms.txt — does your store have an llms.txt file telling AI agents about your brand?

These checks don’t affect individual product scores but surface issues that would prevent any optimization from having effect.

Per-product AI Readability Score

Each product receives a score from 0 to 100 across five dimensions:

DimensionWeightWhat it measures
Description quality30%Word count, factual density, semantic richness, LLM comprehension
Attribute completeness25%Color, size, material, fit, style — present and using standard values
Schema markup20%JSON-LD completeness (Product, Offer, ShippingDetails, ReturnPolicy, etc.)
Variant structure15%ProductGroup schema, consistent option naming across variants
Product identifiers10%GTIN, MPN, EAN, barcode population

See How scoring works for a deeper explanation of each dimension.

Reading your report

Score bands

BandScore rangeWhat it means
AI-Ready90–100AI agents can accurately understand, categorize, and recommend this product
Good70–89Minor gaps; AI agents will generally handle this product correctly
Needs Work40–69Significant gaps; AI agents may misrepresent or skip this product
Critical0–39Severe gaps; AI agents are unlikely to recommend this product

The prioritized fix list

Issues are ranked by impact ÷ effort — the fixes that will move your scores the most with the least work come first. Each issue is labeled:

  • Can fix yourself — something you can address directly in Shopify admin
  • Requires Fix Sprint — remediation that Agentic Prep automates in the paid tier

The AI Agent Preview

The report includes a simulation of how ChatGPT or Perplexity would currently describe your three lowest-scoring products. This is illustrative — it shows the effect of your current data gaps, not a guarantee of ranking outcomes.

What to do after the diagnosis

  • If most products score 70+: You may only need to address the “can fix yourself” items
  • If you have products in the Critical or Needs Work bands: The Fix Sprint is the fastest path to improvement
  • Unsure? Review the prioritized fix list — it tells you exactly what each issue costs you and what fixing it would recover