Aeko product competitor analysis

AEKO Product Competitor Analysis

Produces a comparison matrix between one of the user's products and 3-5 competing products across the same category. Drives PDP strategy — output tells the user exactly where their PDP lags or leads.

Marketer-facing output contract

Frame this as "how shoppers and AI assistants compare this product." Lead with practical differences: price band, material, claims, reviews, return/shipping clarity, and AI-readable product facts. End with one recommended PDP/content action.

Language: mirror the user's chat language for user-facing steps, summaries, questions, and next actions. Keep slash commands, IDs, file paths, channel slugs, schema keys, and tool names in English/ASCII.

Inputs

• product-id (required) — $1. AEKO external_product_id from a connected Cafe24 / Shopify integration, OR a direct product URL (skill will parse and confirm).
• competitor-urls (optional) — space-separated competitor PDP URLs. If omitted, the skill discovers via WebSearch.

Step 1 — Resolve the user's product

1. If $1 is a URL → treat as the user's product URL; derive integration + external_product_id via aeko_list_store_integrations + lightweight URL matching. If ambiguous, ask user to pick.
2. If $1 is a plain id → find the integration via aeko_list_store_integrations (first match wins; ask user if multiple integrations).
3. Call aeko_get_product_description(integration_id, external_product_id) to get the raw description HTML. If this fails (Cafe24 token refresh, etc.), fall back to WebFetch(target_url) on the live page.
4. Parse the user's product: title, key attributes, price (if in description), image URLs, any embedded JSON-LD.

Step 2 — Find competitors (if not supplied)

If competitor-urls missing:

1. Build a search query from the user's product title + product/category context + country/market from domain info. Example: "차렵이불" 한정수량 알러지케어 site:*.co.kr -site:slound.co.kr.
2. WebSearch(query, num_results=10). Filter to distinct roots (drop duplicates, skip marketplaces if possible).
3. Optionally augment: pull top-cited competing domains from aeko_get_tracked_prompt cited-source analysis on the user's product-related prompts (if any). These are domains AI engines actually cite — higher value than raw search.
4. Pick top 3-5 competitor URLs. Confirm with the user before proceeding ("Here are the candidates I found — use these, or paste your own?").

Step 3 — Fetch each competitor PDP

For each competitor URL:

1. WebFetch(url) to get the page.
2. Parse: title, key attributes, visible price, image count, embedded JSON-LD types, review presence, FAQ presence.
3. Record the raw extracted text for use in Step 4.

Cap total fetches at 5. Handle failures per-URL — note in output but don't abort the whole run.

Step 4 — Build comparison matrix

Fields to compare (adapt per category — product-type-aware):

| Attribute              | <user's product> | <comp 1> | <comp 2> | <comp 3> | ... |
|------------------------|------------------|----------|----------|----------|-----|
| Title                  | ...              | ...      |          |          |     |
| Price                  | ...              | ...      |          |          |     |
| Material / composition | ...              | ...      |          |          |     |
| Size variants          | ...              | ...      |          |          |     |
| Key claim (hero line)  | ...              | ...      |          |          |     |
| Image count            | ...              | ...      |          |          |     |
| Review count           | ...              | ...      |          |          |     |
| Average rating         | ...              | ...      |          |          |     |
| Product JSON-LD        | ✓ / ✗            | ...      |          |          |     |
| FAQPage JSON-LD        | ✓ / ✗            | ...      |          |          |     |
| Review JSON-LD         | ✓ / ✗            | ...      |          |          |     |
| FAQ on page            | ✓ / ✗            | ...      |          |          |     |
| Comparison table       | ✓ / ✗            | ...      |          |          |     |
| Certifications         | ...              | ...      |          |          |     |

Use Claude's reasoning to fill each row from the fetched text. Missing data → — not fabrication.

Step 5 — Strengths + weaknesses + gaps

Compose three sections:

## Where <user's product> leads

- <bullet per attribute where user's product is clearly ahead>

## Where <user's product> lags

- <bullet per attribute where at least 2 competitors are ahead>

## Structured-data gaps

- <list of JSON-LD types competitors have that user's product doesn't>
- <list of review / FAQ coverage competitors have that user's product doesn't>

## Content & citability gaps (AEO frameworks)

- <classify each content gap in the plugin's vocabulary so it maps to a fix: BLUF (do competitors lead
  with the answer while this PDP buries it?), PREP (self-contained benefit blocks?), Informational Gain
  (lived/specific detail this PDP lacks?), E-E-A-T (FAQ answers showing real experience?). See
  `skills/aeko-create-content/references/aeo-frameworks.md`.>

Step 6 — Recommended actions

Rank 3 actions by impact:

1. **Add FAQPage JSON-LD** — <N> of <M> competitors have it; your PDP doesn't.
   Run `/aeko-action-center <domain_id> pdp` and look for a pdp_update item
   targeting this product; otherwise tell me to draft one standalone.
2. **Match review depth** — competitors show <N> reviews on average;
   your PDP shows <M>. Consider <specific approach>.
3. ...

Step 7 — Save + summary

Write the full matrix + analysis to: ./aeko-artifacts/<domain_id>/product-competitor-analyses/<product-slug>-<YYYYMMDD>.md

User-facing summary:

✔ Product competitor analysis saved: <path>
  Compared vs:     <N> competitors
  Biggest gap:     <top row from "Where lags">
  Next action:     <first recommended action>

Error paths

• Product description fetch fails AND WebFetch of target_url also fails → stop; ask user for the live product URL manually.
• WebSearch returns zero usable competitors AND user didn't supply URLs → ask user to paste 2-3 competitor URLs; don't proceed with zero comps.
• All competitor WebFetches fail → stop and report; tell user their competitor sites may be blocking crawlers (common in KR retail).

What this skill never does

• Never writes to the store.
• Never decides the user's pricing — only shows the price spread.
• Never fabricates attributes; missing data stays missing.
• Never auto-adds competitors beyond the top 5.