> AI agents: this is one page from PostHog's docs. Full index of Markdown docs for LLMs: https://posthog.com/llms.txt # Usage and logs Canonical URL: https://posthog.com/docs/endpoints/usage-analytics > **Cloud region:** This page uses US Cloud hosts. For an EU Cloud project, use `https://eu.posthog.com` instead of `https://app.posthog.com` (app host for links). The **Usage** tab on the [Endpoints page](https://app.posthog.com/endpoints) shows analytics across all your endpoints, while the **Logs** tab on an individual endpoint's page shows a log of its executions. Together they help you monitor performance and debug execution issues. ## Usage tab The **Usage** tab aggregates execution data across all your endpoints so you can monitor how they're being used and spot performance issues. ![Endpoints Usage tab](https://res.cloudinary.com/dmukukwp6/image/upload/w_1600,c_limit,q_auto,f_auto/pasted_image_2026_02_06_T13_08_10_081_Z_c4b2450c10.png) ### What you can see - **Overview cards** - Total executions, bytes read, CPU time, average and P95 duration, error rate, and a split between materialized vs direct executions - **Trend charts** - Executions, error rate, query duration, CPU time, and bytes read over time - **Breakdown table** - Per-endpoint stats including executions, bytes read, CPU time, and error rate ### Filtering and breakdowns You can filter by date range, specific endpoints, and execution type (materialized vs direct). Use the breakdown options to slice the data by: - **Endpoint** - Compare usage across different endpoints - **API key** - See which keys are driving the most usage - **Execution type** - Understand the split between materialized and direct executions ## Logs tab The **Logs** tab on an individual endpoint's page shows one entry per execution of that endpoint. Each entry is logged at the `INFO` level on success or `ERROR` on failure, and includes searchable `key=value` metadata: - **path** - Whether the query ran via `inline` (direct) or `materialized` execution - **cache** - Cache outcome (`hit` or `miss`) - **duration\_ms** - How long the execution took in milliseconds - **rows** - Number of result rows returned - **version** - The endpoint version that was executed - **error** - Error type, if the execution failed Rejected requests (invalid parameters, version not found, etc.) also appear as `ERROR` entries so validation failures aren't invisible. Filter logs by level or use free-text search to find specific executions. The `execution_id` returned in the [run response](https://posthog.com/docs/endpoints/execution.md#response-format) maps directly to a log entry, so you can trace any result back to its log.