Problem
aspire.dev has general analytics, but it cannot show where readers abandon important content and UX journeys. Without ordered funnel stages, it is difficult to tell whether improvements are needed in discovery, documentation, commands, navigation, or recovery experiences.
Proposed outcome
Add consent-gated funnel telemetry for these journeys:
- CLI installation
- Getting started / first app
- Search success
- Integration adoption
- Deployment intent
- Troubleshooting recovery
- 404 recovery
- Existing-app adoption
Each funnel should use a stable, versioned event contract with ordered stages and bounded dimensions. It must not collect raw search text, arbitrary URL queries, raw referrers, or a new user identity. Cross-page completion signals should use short-lived same-origin state.
For CLI installation, the reliable sequence ends at command_copied. Server-side script-request measurement is out of scope until StaticHost uses a compatible telemetry destination. Its existing Azure Monitor exporter is configured with the browser 1DS token, while Azure Monitor requires an Application Insights connection string.
Acceptance criteria
- Funnel stages cover entry, meaningful progress, and completion or recovery outcomes that the current telemetry pipeline can reliably observe.
- Shared components and relevant English/Japanese guides emit consistent metadata.
- Search and 404 continuations correlate only same-origin destinations and expire promptly.
- Unknown 404 paths are reported through a fixed
/404/ sentinel.
- Automatic analytics do not leak query-bearing URLs or raw referrers.
- Unit, component, and browser tests cover the event contract and user journeys.
Problem
aspire.dev has general analytics, but it cannot show where readers abandon important content and UX journeys. Without ordered funnel stages, it is difficult to tell whether improvements are needed in discovery, documentation, commands, navigation, or recovery experiences.
Proposed outcome
Add consent-gated funnel telemetry for these journeys:
Each funnel should use a stable, versioned event contract with ordered stages and bounded dimensions. It must not collect raw search text, arbitrary URL queries, raw referrers, or a new user identity. Cross-page completion signals should use short-lived same-origin state.
For CLI installation, the reliable sequence ends at
command_copied. Server-side script-request measurement is out of scope until StaticHost uses a compatible telemetry destination. Its existing Azure Monitor exporter is configured with the browser 1DS token, while Azure Monitor requires an Application Insights connection string.Acceptance criteria
/404/sentinel.