Back to BlogSessions by assistant. See how much of your traffic comes from ChatGPT, Claude and other clients. Unique linked users. Count how many customers have connected their account. Tool calls per session. This shows how deep each conversation goes. Top intents. These are the questions people actually ask, clustered and ranked. Zero-result rate. This tracks how often your connector couldn't help, and why. Latency (p50/p95). Slow tools get skipped by assistants. Error rate by tool. Catch broken integrations before customers notice. Conversions. Bookings, checkouts and sign-ups that started in an assistant. Revenue attribution. These are real numbers to bring to your next board meeting.
// Analytics
You Can't Optimize What You Can't See: Measuring AI Assistant Traffic
Xarvix TeamSeptember 8, 20264 min read
The analytics blind spot
When a customer finds you through ChatGPT, there's no page view, no referrer and no cookie. Traditional analytics simply doesn't see it. Meanwhile, that conversation may be shaping a purchase decision.
The metrics that matter
Reach
Engagement
Performance
Outcomes
Closing the loop
Analytics only pays off when it changes the product. Xarvix Insights connects every metric back to the tool and the prompt behind it, so you know which tool description to rewrite, which endpoint to speed up, and which new tool to build next.