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

# Metrics: Monitor data quality and pipeline health in PRIZM

> Metrics are the continuous measurements PRIZM runs against every asset, source, and system to answer two questions: is this data trustworthy, and is this pipeline healthy?

<script type="application/ld+json">
  {`{
            "@context": "https://schema.org",
            "@type": "TechArticle",
            "headline": "Metrics: Monitor data quality and pipeline health in PRIZM",
            "description": "Metrics are the continuous measurements PRIZM runs against every asset, source, and system to answer two questions: is this data trustworthy, and is this pipeline healthy?",
            "url": "https://docs.dqlabs.ai/concepts/metrics",
            "publisher": {
              "@type": "Organization",
              "name": "DQLabs Inc",
              "logo": "https://media.brand.dev/332adc35-5bc4-4d2b-bf78-256aa4a5e414.svg"
            }
            }`}
</script>

A **Metric** is a single, repeatable measurement PRIZM runs against a piece of your data estate — a table, a column, a warehouse, or two datasets checked against each other. Metrics are what turn a static catalog into a living one: instead of finding out a table stopped updating when a downstream dashboard breaks, PRIZM's metrics catch it the moment it happens.Every asset carries a set of metrics, visible on its **Metric tab** (see [Asset Details](/concepts/assets#asset-details)), and every metric belongs to one of six **contexts** — a lens that determines who the metric is for and what question it answers.

## The six metric contexts

<CardGroup cols={2}>
  <Card title="Operational" icon="tower-broadcast">
    Is this asset arriving on time, at the right volume, with the right structure?
  </Card>

  <Card title="Performance" icon="gauge-high">
    Is the warehouse or system underneath it keeping up?
  </Card>

  <Card title="Structural" icon="shapes">
    What does this column actually look like, value by value?
  </Card>

  <Card title="Business" icon="building">
    Does the data satisfy the rules your business cares about?
  </Card>

  <Card title="Reconciliation" icon="arrows-left-right">
    Does this data agree with the same data somewhere else?
  </Card>

  <Card title="Semantic" icon="diagram-project">
    Rolled up from the underlying asset-level scores, how healthy is this business concept overall?
  </Card>
</CardGroup>

**Operational** metrics (Volume, Freshness, Schema) and **Performance** metrics (Usage, Queries, Credits) are auto-generated the moment an asset or warehouse is connected — no setup required. **Structural** metrics (Distribution, Frequency, Pattern, Statistics) are also automatic: PRIZM profiles every column and generates them for free. **Business** and **Reconciliation** metrics are the two contexts you configure yourself, because they encode judgment PRIZM can't infer on its own — a discount that can't exceed 100%, or a row count that must match between a source system and the warehouse.

<Note>
  See [Metric Overview](/architecture/metric) for the full breakdown of categories and individual metrics within each context.
</Note>

## Two questions every metric answers

Under the hood, every metric is configured to answer one or both of these questions:

* **Should this raise an alert if something looks wrong?** This is a metric's **Monitor** setting. Operational and Performance metrics are always `Monitor: Yes` — they exist specifically to catch problems in real time and feed [Alerts](/concepts/alerts).
* **Should this count toward the asset's overall trust score?** This is a metric's **Score** setting. Structural metrics always contribute to score. Business metrics contribute only for the Custom, Conditional, and Standalone types — Behavioral metrics track trends over time and are excluded from scoring, since a trend deviation isn't the same as a quality defect.

| Context        | Monitor (drives alerts) | Score (drives trust score)                           |
| -------------- | ----------------------- | ---------------------------------------------------- |
| Operational    | Yes                     | No                                                   |
| Performance    | Yes                     | No                                                   |
| Structural     | —                       | Yes                                                  |
| Business       | —                       | Custom, Conditional, Standalone: Yes. Behavioral: No |
| Reconciliation | Yes                     | —                                                    |
| Semantic       | —                       | Rolled up from the underlying asset-level scores     |

This split matters in practice: a warehouse running slow shouldn't drag down a table's data quality score, but it absolutely should trigger an alert. Keeping observability (is something broken right now) separate from scoring (is this asset trustworthy overall) keeps both signals meaningful.

## The Metrics page

Beyond the per-asset Metric tab, PRIZM also rolls all six contexts up into a live, org-wide view on the **Metrics page** — the fastest way to see the health, performance, and business consumption fitness of your entire data and pipeline ecosystem.  Each metric overview and timeline page provides comprehensive information about the metric, including all breaches that have occurred over a given period.. The chart defaults to a 30-day window. Use the date control in the chart header to adjust the range.

<Note>
  Metrics refresh automatically as new data is ingested. To force an immediate update without waiting for the next refresh cycle, click the **refresh icon** in the top-right corner of the Recent Alerts panel.
</Note>

## Bulk threshold configuration and cloning

For Business metrics with manual thresholds, the Metrics page's **More** menu offers **Import** (bulk-upload threshold values — upper/lower bound, condition, and manual threshold — for many metrics at once via file import) and **Bulk Edit** (apply the same change across a selected set of metrics), alongside **Select Columns** and **Download**. Custom Threshold conditions also support the full set of comparison operators (equals, greater than, less than, between, etc.) consistently regardless of whether the underlying field is numeric or text.

Any metric can also be **cloned** from its row-level actions — useful for standing up a variant of an existing metric (for example, the same check scoped to a different lookback window) without rebuilding it from scratch.

## From metric to action

When a metric breaches its threshold, PRIZM raises an [Alert](/concepts/alerts) with a severity level so you know how urgently to respond. If the underlying problem needs a tracked fix rather than a one-off glance, turn that alert into an [Issue](/concepts/issues) and manage it through to resolution.

<CardGroup cols={2}>
  <Card title="Explore every metric" icon="list-check" href="/architecture/metric">
    The full reference of contexts, categories, and individual metrics — including which ones are auto-generated and which you configure yourself.
  </Card>

  <Card title="See metrics on an asset" icon="album-collection" href="/concepts/assets">
    How metrics, alerts, and scores surface together on an individual asset's detail page.
  </Card>
</CardGroup>

Together, the six contexts and the Metrics page give you two complementary views of the same signal: one org-wide dashboard for a quick pulse check, and one detailed model underneath it for understanding exactly why that pulse looks the way it does.