Agentic Commerce Readiness Audit

3/9

score

Strong Foundations, Missing Structured Layer

Crawlability and basic infrastructure work; structured data emission is the primary gap for agent retrieval.

Area breakdown

1. Crawlability

2/3

2. Structured data

1/4

3. Content quality

0/2

The 9 criteria

✓ C1

AI crawler access

All 8 tested AI bots allowed in robots.txt.

✗ C2

PDP server-side render

Only 1/3 sampled PDPs render all core elements in raw HTML.

✓ C3

Sitemap freshness

Sitemap has 261 URLs with fresh <lastmod> timestamps.

✗ C4

Product schema

0/3 PDPs have complete Product JSON-LD.

✗ C5

AggregateRating

0/3 PDPs ship Review or AggregateRating schema.

✗ C6

FAQPage / HowTo schema

No FAQPage or HowTo schema on checked pages.

✓ C7

Organization schema

Organization JSON-LD present with sameAs array (5 profiles).

✗ C8

PDP titles

2/3 sampled titles include differentiator cues.

✗ C9

Description depth

0/3 descriptions answer buyer-context questions.

What needs fixing

The top 9-criterion grid shows all checks. Detail below covers only items that need action.

C2 PDP server-side render Fail

Checked: 3 sampled PDPs via raw HTML fetch
Found: 2 of 3 PDPs don't render all core fields in raw HTML. Missing: description, price, title. Examples: skin-hero-deluxe-mini-5ml-copy, three-ships-cosmetic-bag.
Fix: Audit the product-template rendering path to confirm title/price/description appear in raw HTML without JS. Agents cannot execute JavaScript on merchant sites.

C4 Product schema Fail

Checked: 3 sampled PDPs — raw HTML via source-level fetch
Found: Only 0/3 PDPs ship complete Product JSON-LD with all commonly-recommended fields (name, image, brand, offers.price, offers.availability, and an identifier).
Fix: Ensure every PDP template emits Product JSON-LD with the Google rich-results minimum fields. See Google's product structured-data docs.

C5 AggregateRating Fail

Checked: 3 sampled PDPs — Review / AggregateRating schema
Found: 0/3 PDPs expose rating data as JSON-LD with real ratingValue and reviewCount. Others likely display reviews visually but don't emit structured data — typically a widget toggle that hasn't been enabled.
Fix: Enable review-schema output in your review plugin (Yotpo, Judge.me, Okendo, etc.) — often a single setting. See schema.org/AggregateRating.

C6 FAQPage / HowTo schema Fail

Checked: Homepage and discoverable FAQ / how-to-use / about pages
Found: No FAQPage or HowTo JSON-LD detected. If your store publishes how-to / routine content, it's currently invisible as structured data.
Fix: Wrap existing FAQ content in FAQPage JSON-LD and any routine / tutorial content in HowTo JSON-LD. These map directly to the "how do I use X" queries AI shopping agents handle. schema.org/FAQPage / HowTo.

C8 PDP titles Fail

Checked: 3 PDP titles sampled from the sitemap
Found: Only 2/3 titles include specific use-case, target audience, or differentiator cues. Examples: "Bright Eye Besties Bundle: Illuminating Eye Care", "Jelly Drops Mushroom Hyaluronic Acid I Three Ships".
Fix: Expand generic titles ("Brand Name Face Cream") with specific differentiators ("Retinol Alternative Night Serum for Sensitive Skin"). Titles match the natural-language shopper queries AI agents parse.

C9 Description depth Fail

Checked: 3 sampled PDPs — full description text
Found: 0/3 descriptions cover 3+ of: target skin/fit type, comparison to similar products, ingredient/material rationale, sizing reasoning, usage guidance, counter-indications. Most are spec-heavy with limited buyer context.
Caveat: Evaluation via rendered-response text. Descriptions may include buyer-context content in expandable sections or tabs that weren't surfaced.
Fix: Add a structured "Why you'll love it" block to every PDP — who it's for, how it compares, when to use, what it isn't for. At catalog scale this is a content pipeline (LLM-assisted generation from existing spec sheets is plausible).

View complete check log (all 9 criteria)

✓ C1 AI crawler access. robots.txt allows all 8 tested bots (GPTBot / PerplexityBot / ClaudeBot / anthropic-ai / Google-Extended / Amazonbot / CCBot / FacebookBot).

✗ C2 PDP server-side render. 1/3 sampled PDPs return complete product data without JS. Others missing one or more of title/price/description.

✓ C3 Sitemap freshness. Valid sitemap with 261 recently-updated URLs (most recent: 2026-04-17T06:03:54-04:00).

✗ C4 Product schema. 0/3 PDPs ship complete Product JSON-LD. Others missing required fields or the block entirely.

✗ C5 AggregateRating. 0/3 PDPs carry rating schema. Reviews likely displayed visually but not exposed as structured data on others.

✗ C6 FAQPage / HowTo schema. Not present on homepage or auxiliary help pages.

✓ C7 Organization schema. Homepage ships Organization JSON-LD with name, url, and sameAs (5 profiles).

✗ C8 PDP titles. 2/3 sampled titles go beyond brand + generic category.

✗ C9 PDP descriptions. 0/3 sampled descriptions cover 3+ buyer-context cues. Others are spec-heavy with limited context.

Fix priority

Ranked by leverage on agent discoverability, highest first.

Ship complete Product JSON-LD on every PDP
Highest-leverage structured-data change. Ensures AI agents can extract name, price, availability, and identifiers cleanly. Lightweight if product data already exists server-side.
Enable AggregateRating output in your review plugin
Usually a single setting in Yotpo / Judge.me / Okendo that exposes existing review data as JSON-LD. Reviews already live on the site — this just makes them machine-readable.
Wrap FAQ / how-to content in FAQPage or HowTo JSON-LD
Maps directly to "how do I use X" queries AI shopping agents handle. Lightweight if the FAQ template has Q&A as structured data; heavier if mixed with presentational HTML.

Implementing the top fixes would move this brand materially upward on this framework. Final score depends on implementation quality and broader content improvements.

The real story

Three Ships Beauty has foundational gaps that limit agent visibility.

A 3/9 score means multiple layers need attention. Prioritize structured-data basics (C2, C4, C5) before content-level work.