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

# Metrics

> Explore key indicators, build dashboards, and share insights from a single place to understand what's happening across your projects.

The new version of **Metrics** brings the project's analytics into a single place. From [apps.jelou.ai/metrics/v2](https://apps.jelou.ai/metrics/v2) you can review your indicators, build dashboards with the metrics you care about, generate executive insights, and share charts with your team — all under the same global filters.

<Info>
  The **Metrics Administrator** can create, edit, and delete dashboards, insights, and custom metrics. The **view-only** role can open the panel and inspect the data, but can't modify anything. Permissions are assigned from **Settings → User management**.
</Info>

## Overview

The panel is split into three areas:

* **Global filter bar** — date range and filters shared by every chart in the active section.
* **Left sidebar** — switches between **Dashboards** and **Insights**. The **General Summary** dashboard ships preloaded; other dashboards and insights are yours to create.
* **Content** — metric charts you can reorder and resize.

## General Summary

This is the default dashboard. It consolidates the most consulted indicators of the project:

<AccordionGroup>
  <Accordion title="KPI cards">
    Four fixed indicators: **active users**, **workflows**, **conversations**, and **HSM**. Each shows the period value and its change. Selecting one updates the volume chart below to that indicator.
  </Accordion>

  <Accordion title="Volume chart">
    Time series of the selected KPI over the active date range. When the range is longer than 90 days, the series is grouped by month automatically.
  </Accordion>

  <Accordion title="Project user flow">
    Shows how users move across the project's workflows and what share completes or abandons each branch.
  </Accordion>

  <Accordion title="Workflow performance">
    Table with each workflow's performance metrics in the period. A filter lets you include or exclude **workflows shared** with other projects.
  </Accordion>

  <Accordion title="Top words by workflow">
    Table of the most frequent terms, broken down by workflow.
  </Accordion>
</AccordionGroup>

## Global filters

The top bar applies filters to every chart in the active section.

<AccordionGroup>
  <Accordion title="Date range">
    Pick a quick range (today, last 7 days, last 30 days, this month, and more) or define a custom range. The date is always global and doesn't appear in the additional filter picker — it applies to every metric.
  </Accordion>

  <Accordion title="Additional filters">
    From **Filters** you can add or remove dimensions on the global bar: **channel**, **team**, **provider**, **currency**, **environment**, **biometric type**, **criterion**, **workflow**, **node**, and **workflow name**. Each dashboard remembers which filters are promoted to the global bar and which stay available per chart.
  </Accordion>

  <Accordion title="Promote a filter from a chart">
    Each chart exposes its own filters, and you can **promote** any of them to the global bar. Once promoted, the filter moves to the top bar and applies only to the charts that already have it among their dimensions — the rest are unaffected. Below the filter you'll see how many charts on the dashboard it affects.
  </Accordion>

  <Accordion title="Maximum window">
    Some metrics have a day limit to protect performance; when you hit it, the panel shows a hint with the maximum allowed.
  </Accordion>
</AccordionGroup>

<Tip>
  Your browser timezone is sent with each query so daily buckets match your local time.
</Tip>

## Custom dashboards

Beyond the **General Summary**, you can create your own dashboards to group the metrics that matter to a team, channel, or initiative.

<Steps>
  <Step title="Create a dashboard">
    From the left panel, on the **Dashboards** tab, press **+ Create dashboard**. Choose to start blank or from a [template](#available-templates), enter a name, and confirm.
  </Step>

  <Step title="Add metrics">
    Press **Add metric** to open the catalog. Metrics are organized in categories: **Inbox**, **E-commerce**, **Payments**, **Voice**, **Biometrics**, **Brain**, **AI**, and **General**. You can search by name or filter by category.
  </Step>

  <Step title="Organize the dashboard">
    Each dashboard supports up to **10 metrics**. Drag the charts to reorder them and use the corners to resize.
  </Step>

  <Step title="Rename, duplicate, or delete">
    You can **rename** each dashboard, **duplicate** it to start from a copy, or **delete** it permanently.
  </Step>
</Steps>

### Available templates

When you create a dashboard you can start from a template with the essential metrics for an area:

| **Dashboard**             | What it includes                                                                                                                                                    |
| ------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **General Summary**       | System dashboard: active users, workflows, conversations, HSM, volume, user flow, workflow performance, and top words. It can't be renamed, duplicated, or deleted. |
| **Operators and Tickets** | Essential Inbox metrics.                                                                                                                                            |
| **Biometrics**            | Essential biometric verification and KYC metrics.                                                                                                                   |
| **Ecommerce**             | Essential Ecommerce metrics: sales, products, and customers.                                                                                                        |
| **Brain Studio**          | Essential Brain metrics: agent evaluations and most-used words.                                                                                                     |
| **Payments**              | Charges, success rate, providers, and currencies.                                                                                                                   |
| **AI consumption**        | Essential AI token cost and consumption metrics.                                                                                                                    |
| **Outbound Messaging**    | WhatsApp template send performance: delivery KPIs, daily trend, send origin, and funnel.                                                                            |

## Chart types

Catalog metrics are drawn with one of these types. When you create a [custom metric](#custom-metrics) you pick from the subset that source supports.

<AccordionGroup>
  <Accordion title="Indicators">
    * **Number** — a single value compared to the previous period: current value, growth, and an optional breakdown at the foot of the card. Typical use: headline totals — conversations, users, messages sent.
    * **Number group** — several indicators on one card, each with its own format (integer, decimal, percentage, duration, or currency) and its change. Typical use: rates of the same process.
  </Accordion>

  <Accordion title="Time series">
    * **Line** — how a value evolves in the date range, with one point per date.
    * **Area** — the same series, with the area under the curve filled. Typical use: accumulable volumes, such as messages or sessions per day.
    * **Multi-line** — several series on the same plane, with a legend to show or hide each one. Typical use: comparing categories at once (by channel, by status).
  </Accordion>

  <Accordion title="Comparison and composition">
    * **Bars** — vertical bars by category or by date. Typical use: counts with few labels.
    * **Horizontal bars** — ordered ranking; works better when labels are long. Typical use: top workflows, reasons, or agents.
    * **Stacked bars** — each bar splits into segments so you can see the mix of the total. Typical use: sessions by channel per day, HSM statuses by date.
    * **Pie** — percentage split across a few categories. Typical use: share by channel or session type.
  </Accordion>

  <Accordion title="Distribution and intensity">
    * **Histogram** — how a variable falls into contiguous ranges (buckets). Typical use: session duration or response times.
    * **Heatmap** — intensity matrix by color. Typical use: conversations by hour and day of week.
  </Accordion>

  <Accordion title="Flows and funnels">
    * **Funnel** — classic conversion: each level shows how many reached that stage versus the previous one.
    * **Stage funnel** — columns with a percentage, its own denominator, and an alert when a stage falls into a risk zone. Typical use: identity verification or checkout.
    * **Sankey** — flow between nodes proportional to volume, with drop-off points. Typical use: journeys across workflows.
    * **Journey Sankey** — full-width variant for a step-by-step journey.
  </Accordion>

  <Accordion title="Tables, text, and geography">
    * **Table** — text, number, or percentage columns, with status badges, progress bars, filters, and pagination. Typical use: detail by workflow, agent, or template.
    * **Word cloud** — terms sized by frequency. Typical use: most-used words in conversations or searches.
    * **Bubble map** — bubbles whose size is the volume at each location. Typical use: results by city or region.
  </Accordion>
</AccordionGroup>

## Custom metrics

If the catalog doesn't cover your case, you can register your own metrics without leaving the panel.

<Tabs>
  <Tab title="Event-based">
    Turn the events you emit from your workflows into charts. Each metric compares **up to 5 events on the same chart**, labeled A–E. Each event is configured separately with:

    * **Event** you want to measure.
    * **Measure** to compute: total events, unique users, total sessions, or sum of a numeric value.
    * **Filters** by event property.
    * **Breakdowns** to segment the series by property — we recommend no more than 2 breakdowns.

    Available chart types are **Line chart**, **Bar chart**, **Indicator**, **Breakdown tree**, and **Pie chart**. The editor ships a live preview panel with the range you pick.

    To generate the events, open the node in Brain Studio, go to the **Events** tab, and register a snake\_case name plus the properties you want to filter or break down. You can configure up to 10 events per node, each with up to 5 properties.

    <Card title="How to generate events" icon="chart-line" href="/en/guides/observabilidad/eventos">
      Configure tracking events on your nodes to feed custom metrics.
    </Card>
  </Tab>

  <Tab title="Based on Databases">
    Build metrics on top of your own databases without writing SQL. Pick a template:

    * **Total count** — number of records in the collection.
    * **Count by group** — total grouped by a field (channel, status, country…).
    * **Count over time** — daily, weekly, or monthly evolution.
    * **Numeric aggregate** — sum, average, min, max, or distinct count of a field.
    * **Funnel** — step sequence to measure conversion.

    Each template defines which chart types it supports from **Indicator**, **Bar chart**, **Horizontal bars**, **Line chart**, **Pie chart**, **Table**, and **Funnel**. You can add filters with **equal to**, **different than**, **greater than**, and **less than** operators, and choose the metric's own date range if it shouldn't follow the global one.
  </Tab>
</Tabs>

<Warning>
  Deleting a custom metric removes it from every dashboard where it's published. If you need adjustments, use **Edit** instead of recreating it.
</Warning>

## Insights

**Insights** are executive documents generated through the **Jelou Agent** that summarize the project's evolution over a date range. Unlike dashboards, an insight:

* Is stored as rich HTML with **text**, **tables**, and **embedded charts**.
* Can be **printed** or **downloaded** to share outside the platform.
* Keeps the date range it was generated with, so you can compare monthly deliveries.

Insights are **highly customizable**: when asking the Jelou Agent you can tell it how to structure the report, which **chart types** to include, which **color palette** to use, or which sections to highlight. If you want those preferences to always apply, save them to the **agent's memory** and they'll be honored automatically on every insight it generates — including the colors and styles across all charts.

From the **Insights** tab on the left panel you can search, rename, and delete existing insights.

## Sharing charts

Every chart offers three useful actions for team collaboration:

* **Share chart** — creates a public read-only link so anyone with the link can view the chart with its current filters.
* **Use via API** — opens a side drawer with a ready-to-run snippet to consume the metric from your backend or notebook, in the language you pick: **cURL**, **JavaScript**, **Python**, or **PHP**. Each snippet includes the active filters, the timezone, and `companyId`.
* **Configure filters** — pin chart-specific filters, different from the global ones, when the metric supports it.

## Best practices

<AccordionGroup>
  <Accordion title="Start with the General Summary">
    Before creating dashboards, review the **General Summary** to spot which indicators drive decisions and deserve their own dashboard.
  </Accordion>

  <Accordion title="One dashboard per decision">
    The best dashboards answer a single operational question (for example, "how is the Black Friday campaign doing?"). Split into several dashboards when you start mixing very different audiences.
  </Accordion>

  <Accordion title="Document your custom metrics">
    When you create an event- or database-based metric, use a descriptive name — the same one that will appear in dashboards and insights. Avoid internal shorthand that other users can't interpret.
  </Accordion>

  <Accordion title="Share links, not screenshots">
    A chart's **public link** always reflects up-to-date data, while a screenshot goes stale as soon as the filter or range changes.
  </Accordion>
</AccordionGroup>