A benchmark study by Gorgias, a $100 million ARR AI customer experience leader, tested 13 AI vendors on 212 live mid-market ecommerce stores. The study measured how many customer support conversations AI agents could fully resolve without human intervention. The top five AI agents resolved between 64% and 75% of conversations, with the best performing agent resolving about 70%. The median resolution rate across all vendors was 48%, according to saastr.com.
The benchmark involved sending identical customer messages to each AI agent, which then responded based on live store data such as price, policy, and SKU. Responses were scored blindly against 26 binary checks to verify accuracy. The study pooled data from the last four weeks, with conversation counts ranging from 34 to 561 per vendor. Gorgias itself resolved 64% of conversations, while well-known platforms like Zendesk, Intercom, and Klaviyo resolved less than 42%.
This data highlights the varying effectiveness of AI in ecommerce customer support. While the top AI agents can handle about 70% of queries autonomously, many widely used platforms lag behind, resolving fewer than half. The findings underscore the potential for AI to reduce human workload in ecommerce support but also reveal significant room for improvement. The top five vendors averaged a 70% resolution rate, setting a benchmark for the industry.
The benchmark results provide a concrete measure of AI’s current capabilities in ecommerce support, with the top five vendors resolving 64% to 75% of conversations. This data offers ecommerce businesses a clear reference point when evaluating AI support solutions and highlights the gap between leading AI agents and established platforms like Zendesk and Intercom.