# Analytics

**Available in:** Sites, for measured production traffic. Projects use development Logs and diagnostics. Chat and Agent usage is reported through billing Usage rather than Site Analytics.

Site Analytics helps you understand requests reaching a production destination: which paths are used, how responses behave, and whether latency or errors are changing.

Routing measurements describe traffic. They do not automatically identify unique people, sessions, conversions, or customer journeys.

## Check a released route

Open a known page on the active Site, such as its home page, and note the path and time. In the Site’s **Analytics** view, select a period that includes the visit. Under **Requested paths**, find the path and check that its request count increased. Review **Response statuses** for the same period. A successful visit with a count of zero means traffic collection is still unconfirmed.

## Read the signals

| Measure | Useful question | Interpretation limit |
| --- | --- | --- |
| Latency | Are responses slowing down? | The cause may be code, capacity, or a dependency. |
| Paths | Which routes received activity? | A route hit does not prove a completed workflow. |
| Requests | How much traffic reached the Site? | Requests are not unique visitors. |
| Response status | Are errors increasing? | Investigate individual failures through Logs. |

Compare an appropriate time range with release and operational changes. Treat missing measurements as unavailable rather than zero.

## Measure product outcomes separately

When you need signup completion, feature use, or a conversion funnel, define the application events and their meaning. Supported provider integrations and application instrumentation can supply richer analytics, with their own configuration and privacy requirements.

For example, a successful response from a signup page is not the same as a completed registration. An application event should represent the actual outcome you want to measure.

## Keep the scope clear

Site Analytics belongs to production. Project request observations support development Logs and diagnostics rather than a separate Project Analytics dashboard. Each Site retains its own production access and configuration even when several Sites share a Project's source.

Use [Monitoring](/docs/services/monitoring) for exceptions, [Logs](/docs/services/logs) for investigation, and [Usage](/docs/billing/usage) for billed consumption. Traffic volume and billed usage answer different questions.

## Example: investigate a slower route

A route receives similar traffic but its measured latency increases after an update. Compare the relevant period and release, then inspect Logs and dependencies for the cause. Higher latency alone does not establish that more compute is the right fix.

For signup conversion, define a separate application event representing completed registration. Review [Monitoring](/docs/services/monitoring) for failures and [Usage](/docs/billing/usage) for costs.

## Choose a comparison

Compare the same route and similar time windows when evaluating an update. For example, compare the request board's save endpoint before and after a release, then use Logs to investigate changed errors. Keep request volume separate from completed saves; the application must define that outcome if you want to measure it.