Building a first-party AI agent is one thing, tracking it properly is another. In this clip from our session at Superweek, Jon Su walks through tracking challenges that every analytics team needs to prepare for:
1. The Identity Crisi: LLMs deal in unstructured text, but traditional digital analytics is built on structured data. These two worlds are fundamentally at odds.
2. Metrics Break Down: Page views, clicks, and sessions don't map neatly onto agentic experiences. New metrics like "agentic sessions started" and "agent-assisted conversions" need to replace them.
3. The Cold Start Problem: Agents have no memory between sessions. Even if a user spent an hour on your site, the agent greets them like a stranger every time.
4. PII Everywhere: Users trust agents more than humans and often share sensitive information without thinking. That data can end up inside your analytics pipeline.
5. The Model Changes Beneath You: If you're building on OpenAI, Anthropic, or Google models, vendor-side changes can silently alter your customer experience, like an SEO algorithm update you never heard about.
🎥 Watch the full session: https://snowplow.io/events/track-ai-agents-in-2026