> ## Documentation Index
> Fetch the complete documentation index at: https://www.integrate.io/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# ETL: Viewing Job Insights

> Use the Insights tab to see record counts and runtime per component for an Integrate.io ETL job, plus a timeline that shows where the time went.

The **Insights** tab in the job details window shows how many records each component handled and how long each part of the run took. Use it to find slow steps, confirm row counts at sources and destinations, and see where a failed or stopped run ended.

The Insights tab is available for dataflow package jobs. It is not shown for workflow package jobs.

<Note>
  Job insights is enabled per account and is off by default. If you do not see the **Insights** tab in the job details window, contact Integrate.io support to enable it for your account.
</Note>

**To view job insights:**

<Steps>
  <Step>
    In the jobs list, click **Details** for any job. You can also click **View outputs** or **View errors**.
  </Step>

  <Step>
    The job details window appears. Click the **Insights** tab.
  </Step>
</Steps>

If the job doesn't have outputs, responses, tasks, or errors to show, the window opens on the **Insights** tab.

While a job is still running, the tab displays **Updating while the run is in progress** and refreshes the counts and runtimes as the job runs.

## Timeline

The **Timeline** section shows where the time went during the run. Each bar runs from when that part of the run started to when it finished. Bars that overlap ran at the same time. The total runtime appears next to the section title.

The timeline also marks two zones that aren't part of any step:

* **job starting up**: Before the first step runs, the engine starts up, compiles the package, and submits it to the cluster. This time depends on how many components the package has. It does not increase with the amount of data you process.
* **job finishing up**: After the last step, the engine finalizes the outputs and marks the run complete. This usually takes a few seconds. On a failed run it takes longer while the remaining work is stopped. This zone appears only after the run has ended.

## Components table

The **Components** table lists every component in the package except sticky notes, which process no data. The table has these columns:

| Column | Description |
| - | - |
| Component | The component name and icon. Connector components show the vendor logo. |
| Type | **Source**, **Transformation**, or **Destination**. |
| Records | The number of records the component handled, labeled **read** for sources, **written** for destinations, and **out** for other components. |
| Runtime | How long the component ran. |

For packages with more than one flow, the table groups components by flow (**Flow 1**, **Flow 2**, and so on). Click a flow header to collapse or expand it. A failed flow is highlighted.

### Record counts

Depending on the run, the engine counts records for each component or only where data enters and leaves the run, at sources and destinations. When a component's records weren't counted, the **Records** column shows **not measured**. Hover over a count to see how it was measured.

If a step failed, its components show no counts. The engine doesn't report counts for failed steps.

### Runtime for components that ran together

The engine often runs several components together in a single step. Their time can't be split between them. Components that ran together are grouped in the table, marked **ran together**, and share one runtime for the group. A component that ran across several steps shows the number of steps next to its runtime.

If a run ended before a step completed, for example because the job failed or was stopped, the runtime shows **didn't finish**. The time for that step runs to the end of the run.

## Related

* [Monitoring Jobs](/docs/etl/monitoring-jobs)
* [Viewing Completed Task Outputs](/docs/etl/viewing-completed-tasks-outputs)
* [Viewing Job Error Details](/docs/etl/viewing-job-error-details-for-failed-jobs)


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.