Protocol Compliance Scoring for AI Commerce
> Protocol compliance scoring measures whether a product's data layer meets the field requirements of ACP and UCP — the two primary AI commerce protocols — and is the highest-weighted dimension in FeedBridge's AI Readiness Score at 30%, because protocol gaps directly prevent AI-assisted purchasing regardless of how complete the product's other data is.
---
What Is Protocol Compliance Scoring?
Protocol compliance scoring is the evaluation of a product's data against the documented field requirements of the AI commerce protocols that AI shopping agents use to discover, evaluate, and transact on products. In FeedBridge's AI Readiness Score model, protocol compliance is one of four scored dimensions — and the highest-weighted one at 30%.
The reason for its prominence is straightforward: protocol fields are gate conditions, not optimisation opportunities. A product that is missing trust signal fields cannot support ACP Instant Checkout regardless of how good its title and description are. A product without a GTIN cannot be reliably looked up via the UCP Catalog Lookup API. Protocol compliance gaps are not soft quality issues — they are hard blockers that prevent specific AI commerce capabilities from working on that product.
Protocol compliance scoring in FeedBridge evaluates two protocols in parallel: ACP (Agentic Commerce Protocol) and UCP (Universal Commerce Protocol). Each has its own set of required and recommended field standards, and a product's protocol compliance score reflects how completely it meets both.
---
ACP Compliance Signals
ACP compliance scoring evaluates the fields that the Agentic Commerce Protocol requires for a product to participate in AI-assisted purchasing via ACP-enabled surfaces like ChatGPT Shopping.
Trust signal fields — highest priority: The three ACP trust signal fields are the most critical protocol compliance signals:
- `accepts_returns` (boolean) — whether the merchant accepts returns for this product
- `return_deadline_days` (integer) — the number of days the buyer has to return the product
- `is_digital` (boolean) — whether the product is a digital (non-physical) good
Structured variants — `variant_dict`: For products with multiple options (size, colour, storage, etc.), the `variant_dict` field must be populated as a structured key-value mapping. Products where variant information is embedded in free-text descriptions — rather than structured in `variant_dict` — fail this compliance vector because agents cannot filter by variant programmatically.
JSON-LD feed format: ACP compliance requires that the product feed be in JSON-LD format. CSV or XML feeds without proper JSON-LD structure fail the format compliance vector, which affects checkout session accuracy when the agent constructs a `POST /checkout_sessions` request using the product record.
Q&A pairs as JSON array: The `q_and_a` field must be formatted as a JSON array of objects, not as free text. A product with Q&A content in an unstructured format partially satisfies the enrichment requirement but fails the structural compliance vector, because agents cannot parse unstructured Q&A as machine-readable input.
Structured reviews: The `reviews` field must be a JSON array of review objects. Aggregate star ratings alone do not satisfy this compliance signal; structured review objects with reviewer name, rating, and review text are required.
Sale pricing fields: Where a product is on promotion, `sale_price`, `sale_price_start_date`, and `sale_price_end_date` must all be present and correctly formatted (ISO 8601 for dates). A sale price without valid date bounds fails the sale pricing compliance vector.
---
UCP Compliance Signals
UCP compliance scoring evaluates the fields that the Universal Commerce Protocol requires for a product to participate in AI-assisted purchasing via UCP-enabled surfaces like Google AI Mode and Gemini.
`native_commerce` attribute: The `native_commerce: true` attribute in the Google Merchant Center feed signals that a product is eligible for UCP-powered AI-assisted purchasing. Products without this attribute are not eligible for the UCP checkout experience on Google surfaces, even if all other fields are present.
Product identity — GTIN and MPN: GTIN (Global Trade Item Number) and MPN (Manufacturer Part Number) are the primary product identity signals for UCP Catalog Lookup. The UCP Catalog Lookup API can retrieve products by `item_id` or by barcode (GTIN/MPN). Products without GTIN or MPN cannot be looked up by barcode, reducing the precision with which agents can identify and retrieve them.
Taxonomy path: UCP requires a correctly structured taxonomy category path for products. The path must resolve to a valid leaf node in the taxonomy — not a top-level category like "Electronics" but a specific path like "Electronics > Headphones > Noise-Cancelling." Products with missing or top-level-only taxonomy paths fail this compliance vector.
Availability status: Availability must be explicitly declared as `in_stock` or `out_of_stock`. Products without explicit availability status fail this compliance vector — an undeclared availability is not the same as in-stock, and agents cannot assume availability.
Brand field: The brand field must be populated for UCP compliance. Products without a declared brand fail this vector and may not appear in branded product queries.
---
How Protocol Compliance Scoring Is Presented
In FeedBridge, protocol compliance is scored at the per-product level and rolled up into the overall AI Readiness Score. Each product receives a Protocol Compliance sub-score alongside scores for the other three dimensions. The sub-score contributes 30% to the overall 0–100 score.
The actionable fix suggestions generated by FeedBridge identify compliance failures specifically — distinguishing protocol compliance gaps from content quality or enrichment gaps. A product whose Protocol Compliance sub-score is dragging down an otherwise high overall score will receive fix suggestions targeted at the specific missing trust signals, variant structure, or taxonomy path that is causing the compliance gap.
The color-coded tier system (Excellent, Good, Needs Work, Poor) is applied at both the overall score and sub-dimension level, giving merchants an at-a-glance view of where their most critical compliance gaps are concentrated.
---
Why It Matters for Merchants
Protocol compliance is the prerequisite layer for everything else in AI commerce. A merchant who invests heavily in AI enrichment — adding intent tags, persona arrays, Q&A pairs — but has not addressed trust signals and `variant_dict` will have well-described products that cannot complete an ACP checkout. The enrichment investment pays off only after the compliance foundation is in place.
For merchants preparing to submit their feeds for ACP certification or UCP Merchant Center onboarding, protocol compliance scoring serves as a pre-certification readiness check. The vectors in FeedBridge's protocol compliance dimension correspond directly to the schema fields that ACP conformance testing and Google's UCP integration approval will validate. A product that scores Excellent on protocol compliance in FeedBridge has met the data-layer requirements that those external validation processes check.
Merchants with large catalogs benefit especially from per-product protocol compliance scoring, because compliance gaps are rarely uniform — they tend to be concentrated in specific product categories, recently added products, or products imported from sources with different field standards.
---
FeedBridge Relevance
FeedBridge's AI Readiness Score includes protocol compliance as its highest-weighted dimension. The platform's ACP feed pipeline generates and validates all documented ACP protocol fields — trust signals, `variant_dict`, JSON-array `q_and_a`, structured reviews, and sale pricing — and the UCP feed pipeline validates `native_commerce` eligibility, GTIN/MPN presence, taxonomy paths, availability, and brand fields. Protocol compliance gaps are flagged in the per-product fix suggestions with one-click navigation to the specific field.
The Protocol Compliance sub-score is available per product in the FeedBridge dashboard alongside the other three dimension scores, giving merchants a precise view of which products are blocked from AI-assisted purchasing at the data layer versus which are simply under-enriched.
---
Frequently Asked Questions
Q: If a product passes protocol compliance scoring in FeedBridge, does it automatically pass ACP certification? A: No. FeedBridge's protocol compliance scoring validates the product data layer — field presence, format, and structural correctness. ACP certification validates the merchant's checkout API endpoints, error handling, rate limiting, and webhook delivery. Both layers must be correct for full ACP production readiness, but they are separate checks.
Q: Which protocol compliance signals are most commonly missing in merchant catalogs? A: Based on the field requirements, the ACP trust signals (`accepts_returns`, `return_deadline_days`, `is_digital`) are frequently absent from catalogs that were built before ACP standards were published — they are ACP-specific fields not found in legacy feed formats. GTIN is also commonly missing for private-label products that do not have assigned barcodes.
Q: Can a product score well on protocol compliance if it is in a CSV feed instead of JSON-LD? A: CSV is a valid feed format for discovery and evaluation in some AI surfaces (Google Merchant Center accepts CSV). For full ACP compliance — specifically the nested fields like `variant_dict`, `q_and_a` arrays, and structured reviews — JSON-LD is the recommended format. FeedBridge generates ACP feeds in JSON-LD to meet the structural requirements.
Q: Is `native_commerce: true` automatically added to all products in FeedBridge? A: The `native_commerce` attribute is a product-level flag in the Google Merchant Center export. Merchants control which products receive this flag based on which are ready for UCP-powered purchasing. FeedBridge supports this as a configurable field in the Merchant Center CSV export.
Q: How do I fix a failing `variant_dict` compliance vector? A: In FeedBridge, the fix suggestion for a failing `variant_dict` vector will link directly to the product detail modal, where the structured variant mapping can be added or reviewed. The Universal AI Engine can generate `variant_dict` mappings from unstructured variant text as part of the AI enrichment process.
---
Related Topics
Parent hub: AI Commerce Readiness Scoring
Related concepts:
- AI Readiness Score: What It Measures
- Content Quality Scoring for Product Discoverability
- What Is ACP?
- What Is UCP?
- FeedBridge AI Readiness Score Features
Breadcrumb:
---
Source Documentation
| Claim | Source | Source Class | Reference | |---|---|---|---| | Protocol Compliance 30% of AI Readiness Score | FeedBridge Platform Capabilities April 2026 v2.0 | T1 – FeedBridge Internal | FeedBridge-Platform-Capabilities-April2026.md | | ACP trust signals: accepts_returns, return_deadline_days, is_digital | FeedBridge Platform Capabilities April 2026 v2.0 / OpenAI ACP Spec | T1 – FeedBridge Internal / T1 – Official ACP Docs | FeedBridge-Platform-Capabilities-April2026.md | | ACP variant_dict, q_and_a JSON array, reviews JSON array, sale price fields | FeedBridge Platform Capabilities April 2026 v2.0 | T1 – FeedBridge Internal | FeedBridge-Platform-Capabilities-April2026.md | | UCP `native_commerce` attribute | almcorp.com / Google Merchant Center | T2 – Platform Partner Docs | https://almcorp.com/blog/google-universal-commerce-protocol-ucp-explained/ | | GTIN/MPN for UCP Catalog Lookup by barcode | UCP Spec / FeedBridge Platform Capabilities | T1 – Official UCP Docs / T1 – FeedBridge Internal | FeedBridge-Platform-Capabilities-April2026.md |