---
Last week I pointed a scanner I built at allbirds.com to answer a simple question: does this store look "clear and recommendable" or "a blurry mess" to ChatGPT and Claude?
Result: 93/100. Platform detected (Shopify), llms.txt present, AI crawlers allowed, product feed found, ProductGroup JSON-LD valid, feed/page prices consistent. Gaps: no review aggregation markup, no Organization schema.
Here's the 8-item checklist the scanner runs — you can do all of it by hand in about 10 minutes.
1. robots.txt vs AI crawlers (1 min)
Open yourdomain.com/robots.txt. Look for Disallow targeting GPTBot, OAI-SearchBot, ClaudeBot, or PerplexityBot. Pass: none blocked.
2. llms.txt (1 min, bonus)
Open yourdomain.com/llms.txt. Exists and accurate? Nice. Missing? Don't stress — it's a community convention, not a standard. Fix the rest first.
3. Structured data (2 min)
Paste any product URL into Google Rich Results Test (free). Pass: Product + Offer (price + availability) detected. Review / Organization are bonus.
4. Product data completeness (3 min, spot-check 3 bestsellers)
- [ ] Title has material / dimensions / use case
- [ ] Description is full sentences, not "HOT SALE!!!"
- [ ] Attributes filled (color, size, material…)
- [ ] Image alt text: 8–15 real words, not "IMG_001"
- [ ] Category at the deepest level
- [ ] 8–15 specific tags
Pass: 2 of 3 products clear all six.
5. Feed ↔ page price consistency (1 min)
Compare your feed price against the page price. One says $49, the other $39 → the model flags a conflict and you can get demoted. Pass: identical everywhere.
6. Zombie products (1 min)
Filter your catalog for "out of stock but listed," "obviously mispriced," or "untouched 12+ months." Delist or mark restocking. AI's nightmare is recommending something unbuyable.
7. Reviews + policy pages (1 min)
Bestsellers need real written reviews (≥5, not just stars). Return/shipping pages must exist and read clearly. Pass: both true.
8. Are you in the feed pool? (1 min)
Shopify/Etsy: auto-synced since March 2026, you're almost certainly in. Everyone else: ChatGPT feeds need an OpenAI application (waitlist) or Salesforce/Stripe ACP connection — US-only for now.
Scoring
- 7–8: solid — next battle is product strength and reviews
- 4–6: clear weak spots, fixable in ~2 weeks
- 0–3: unreadable to AI — but it's all data-organizing work, costs time not money
I automated this into a free one-minute scan — same checks, same scoring, no signup:
https://muse.ai/s/free-ai-visibility-checker-qxd5xhdxmnxbbww?utm_source=devto&utm_medium=social&utm_campaign=2026-09-merchant-tutorials
Top comments (0)