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

# Analytics

> Understand how your recommendation engine is performing and use data to guide configuration decisions.

The Analytics page gives you a time-series view of recommendation activity and user engagement. You can explore pre-built metrics, build custom charts, and drill down to individual requests or sessions.

***

## Time controls

### Preset ranges

Click the date button in the toolbar to choose a preset range:

| Preset     | Window                  |
| ---------- | ----------------------- |
| 24h        | Last 24 hours           |
| 7d         | Last 7 days             |
| 30d        | Last 30 days            |
| 90d        | Last 90 days            |
| 6m         | Last 6 months           |
| 12m        | Last 12 months          |
| YTD        | Year to date            |
| This week  | Current calendar week   |
| Last week  | Previous calendar week  |
| This month | Current calendar month  |
| Last month | Previous calendar month |
| Last year  | Previous calendar year  |

### Custom range

Select **Custom** from the date dropdown and enter a start and end date. Click **Apply** to load that window, or **Clear** to return to the previous preset.

### Back and forward

Use the arrow buttons to shift the current window backwards or forwards by one period. For example, if you are viewing "Last 7 days", clicking back loads the 7 days before that.

### Granularity

The **Granularity** dropdown controls how data is bucketed along the time axis: Minute, Hour, Day, Week, Month, Quarter, or Year. The system automatically coarsens the granularity if the selected window would produce more than 1,500 data points.

***

## Context filter

Use the **Context** dropdown to restrict all metrics and charts to a specific context. Select **All contexts** to see aggregate data across your entire account.

***

## KPI cards

The row of cards at the top of the page shows totals for the current window alongside a percentage change versus the equivalent previous period.

* **Recommendations Served**: total number of recommendation responses returned by the API.
* One card per event type you have defined, showing the total count of that event in the current window.

A green value indicates an increase versus the previous period. A red value indicates a decrease. A value of 0% means the metric was unchanged or there is no previous data to compare against.

***

## Custom charts

The widget grid lets you compose your own charts from any available metric.

### Add a chart

<Steps>
  <Step title="Open the widget builder">
    Click **Add chart** in the top toolbar.
  </Step>

  <Step title="Choose a chart type">
    In the widget builder, choose a chart type: Line, Bar, or Area.
  </Step>

  <Step title="Select a metric">
    Select the metric to plot from the dropdown. Available options are **Served** plus one entry per event type.
  </Step>

  <Step title="Add an optional label">
    Optionally enter a custom label. If you leave it blank the metric name is used.
  </Step>

  <Step title="Add the chart">
    Click **Add**.
  </Step>
</Steps>

The chart appears in the grid and displays data for the current time window and filters.

### Remove a chart

Click the close button in the top-right corner of any chart widget to remove it.

### Reset layout

Click the reset icon in the toolbar to remove all widgets and restore the default layout, which shows a single Recommendations Served line chart.

***

## Advanced filters

The **Advanced** dropdown lets you narrow all data to a specific request, session, or user.

| Filter     | What it does                                                                            |
| ---------- | --------------------------------------------------------------------------------------- |
| Request ID | Shows only the served recommendation matching that UUID and any events attributed to it |
| Session ID | Shows only served recommendations and events within a single user session               |
| User ID    | Shows only events from a specific user                                                  |

When any advanced filter is active, the toolbar shows an **On** badge next to the Advanced button. Clear individual fields and click **Apply** to remove filters.

These filters are useful for debugging a specific recommendation request or tracing a single user's journey through your product.

***

## Interpreting results

A few patterns to look for:

**High serve count, low click events**: the engine is returning results but users are not engaging. Review your context filters and ranking rules. You may be filtering too aggressively or promoting items that do not resonate.

**Spikes in a specific event type**: cross-reference with product changes or campaigns deployed around the same time. The session filter can help isolate whether a spike comes from real user behaviour or test traffic.

**Delta showing a large drop after a model change**: compare the current model's performance against the previous window, which pre-dates the training run. If engagement dropped, consider reverting the model from the Training Jobs page while you investigate.