Policy Quality
9.1 Why AI Agents Care About Your Policy Pages
When an AI agent recommends a merchant, it needs assurance that consumer rights are protected. Policy pages are a core input for AI agents evaluating merchant transparency (the T dimension). A vague statement like “all sales are final” and a detailed explanation like “full refund within 30 days of receipt for unused items with tags attached” represent entirely different trust levels from an AI agent’s perspective.9.2 Privacy Policy
Required Content
| Item | Description | Why It Matters |
|---|---|---|
| Data types collected | Explicitly list what personal data you collect | AI agents check whether data collection scope is disclosed |
| Purpose of data use | Explain what collected data is used for | Vague statements like “to improve services” are insufficient |
| Third-party sharing | Whether data is shared with third parties, and with whom | Failure to disclose third-party sharing results in a score deduction |
| Data storage and protection | Where data is stored and how it is protected | Demonstrates security awareness |
| User rights | How users can view, modify, or delete their data | Required by GDPR / CCPA / similar regulations |
| Cookie usage | What cookies are used and how to manage them | Undisclosed cookie usage results in a score deduction |
| Contact information | Contact channel for data protection inquiries | There must be a reachable contact method |
| Last updated date | When the policy was last revised | Outdated policies raise credibility concerns |
Writing Principles
- Be specific, not generic — “We collect your name, email address, shipping address, and payment information” is better than “We collect necessary personal information”
- Use layered structure — Organize with headings and lists; do not write a single continuous essay
- Use plain language — Avoid excessive legal jargon; an ordinary user should be able to understand it
- Display the update date prominently — Show the last updated date at the top of the page
AI Prompt: Generate a Privacy Policy
9.3 Return / Refund Policy
Required Content
| Item | Description |
|---|---|
| Return window | Number of days after receipt within which returns are accepted |
| Return conditions | What condition the product must be in (unused / original packaging / etc.) |
| Non-returnable items | Which products are not eligible for return (custom items / perishables / etc.) |
| Return process | How a user initiates a return (contact support / submit online / etc.) |
| Refund method | Original payment method / store credit / bank transfer |
| Refund timeline | How long until the refund is processed |
| Shipping costs | Who bears the cost of return shipping |
| Exchange policy | Whether exchanges are supported and how to request one |
Writing Principles
- Make conditions explicit — Consumers should be able to determine return eligibility before purchasing
- Make the process clear — Document every step of the return process
- No hidden traps — Do not bury hidden conditions in the policy
- Be specific about timelines — Use concrete numbers of days and dates; do not say “as soon as possible”
9.4 Terms of Service
While terms of service carry less weight than privacy and return policies in AI trust assessments, a complete terms of service page contributes positively to the T dimension. Key content:- Conditions and restrictions of use
- Intellectual property declarations
- Limitation of liability
- Dispute resolution mechanisms
- Governing law and jurisdiction
9.5 Technical Requirements for Policy Pages
- Dedicated URLs — Each policy should have its own URL (
/privacy-policy,/return-policy); do not place them all on a single page - Crawlable — Ensure robots.txt is not blocking policy pages
- Visible in navigation — Link to all policy pages from the page footer or navigation bar
- Schema.org markup — Optional but beneficial; add WebPage markup to policy pages
Next chapter: Business Identity — LEI registration, business information verification, and domain credibility