GEO measurement stack
Three layers of AI-search measurement — and the discipline to label what each one can actually claim.
client analytics work; properties and report-suite specifics withheld.
The measurement framework underneath the GEO work: Layer 1 is clickthrough — AI-referred traffic measured in the site's analytics platform, accessed programmatically through its MCP server with a playbook encoding the full agent workflow. Layer 2 is visibility — how the AI engines describe and cite the brand. Layer 3 is crawler readability. Built explicitly to show stakeholders that a low crawler-optimizer score measures accessibility, not AI-driven business outcomes.
The forensics mattered: the analytics instance had a decoy — a production-labeled report suite that was empty, while the real traffic lived in one with a misleading name. Finding the right source of truth was the first deliverable.
Full case study in progress — artifacts pending clearance or writing time. The insight below is already earned.
A crawler-readability score isn't a business outcome. Build the measurement stack in layers and be precise about what each layer can claim — that precision is the deliverable.
the insight