Benchmark protocol v0.1

Status

Protocol v0.1 — results forthcoming. We publish the protocol before results so methods can be reviewed and replicated.

What we measure

1. AI readiness (audit dimensions: readability, structure, consistency, freshness). 2. AI visibility (mention rate, citation share, share of voice, source mix). 3. Performance (Core Web Vitals / Lighthouse, where available).

Sampling

• Shopify and WooCommerce stores, opt-in;
• stratified by catalog size and category;
• no client-identifying data in published aggregates.

Metrics (definitions)

• mention rate = runs mentioning the brand / total runs;
• citation share = runs citing the store URL / total runs;
• share of voice = brand mentions / (brand + competitor mentions);
• median values across repeated runs; report N and time window.

Publication policy

• publish methods and aggregate results;
• never publish client-identifying data;
• report limitations and conflicts;
• no pay-to-include.

Replication

Evidence schemas, prompt sets and aggregation scripts are documented. We will publish a replication package with the first results.