Benchmark

The average site scores 29/100 for agent product-findability

Across 153 sites run through the Serge scanner, the average agent product-findability score is 29/100 and the median is 26/100. In plain terms: on most sites, an AI agent may not be able to find a product and add it to the cart.

What this means

A score of 29/100 is the gap between a sale and a silent loss. When a customer asks Claude, ChatGPT, or Operator to buy something, the agent either completes the task or gives up — and the failure never shows up in GA4. Most sites in this cohort score low because their product data, navigation, or cart were not built for a non-human shopper. The score is a measurement, not a verdict: every failed check ships with a specific fix.

Methodology

• Cohort: the 153 distinct domains run through the Serge scanner to date. No site is named, ranked, or identified — only aggregate figures are published.
• Self-selection: this cohort is sites whose owners chose to run Serge, which skews toward stores suspected of having product-findability problems. Read the figures as a directional baseline, not a population average of all e-commerce.
• Metric: a deterministic 0-100 product-findability score from a crawler plus static analysis (structured data, navigation, semantic HTML, cart and form accessibility, bot posture). No AI runs at scan time.
• Sampling: the most recent scan per domain, so each site is counted once.
• Privacy: figures are aggregate only. An individual site score is never published here.

Cite this benchmark

Serge AI-Agent Product-Findability Benchmark (2026): across 153 sites scanned, the average score was 29/100. Source: serge.ai/benchmark