> ## 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.

# Criticality Scoring

> Know instantly which assets matter most, before they break.

<script type="application/ld+json">
  {`{
            "@context": "https://schema.org",
            "@type": "TechArticle",
            "headline": "Criticality Scoring",
            "description": "Know instantly which assets matter most, before they break.",
            "url": "https://docs.dqlabs.ai/architecture/criticality",
            "publisher": {
              "@type": "Organization",
              "name": "DQLabs Inc",
              "logo": "https://media.brand.dev/332adc35-5bc4-4d2b-bf78-256aa4a5e414.svg"
            }
            }`}
</script>

**Criticality** indicates how important an asset is to the business — used to prioritize monitoring, alerting, and governance effort toward the assets that matter most.

| Mode       | Behavior                                                                                                                                                                                                                                            |
| :--------- | :-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Auto**   | Prizm dynamically calculates criticality based on signals like downstream lineage impact (how many assets depend on it), usage frequency (queries/views), and dependency count. The score updates automatically as these signals change over time.  |
| **Manual** | A user with appropriate permissions overrides the system-calculated value to reflect business judgment — for example, marking a low-usage but compliance-critical table as "Critical" even though usage signals alone wouldn't surface it that way. |

<Tip>
  Use **Manual** mode when business context outweighs system-derived signals — for example, a regulatory reporting table that's rarely queried but mission-critical.
</Tip>

***

### How Auto Criticality Is Derived

**Criticality** is an automatically derived score representing how important an asset is to the business, based on five factors:

**Usage** carries the most weight, combining how many unique users access the asset with how many read queries it receives — assets used by many people, frequently, score higher.

**Dependency** is equally weighted with Usage, capturing how many upstream and downstream assets rely on it — the more connected an asset is, the bigger the impact if it breaks.

**Metadata** completeness contributes a moderate amount, driven mainly by whether the asset has linked business terms and domain/app/product classification, with smaller contributions from tags (whether DQLabs-detected or sourced) and a written description — well-governed assets are treated as more deliberately important.

**Freshness** and **Attribute Criticality** play a smaller, supporting role — freshness reflects how recently the asset was updated, while attribute criticality reflects whether the asset contains sensitive or business-critical columns.

**Volume** has the smallest influence, based simply on row count — larger datasets nudge the score up slightly, but size alone isn't a strong indicator of importance.

Together, **Usage and Dependency dominate** the score, which is exactly why low-usage but compliance-critical assets (like a rarely-queried regulatory table) often need a **Manual override** — the automatic calculation won't fully capture business-mandated importance on its own. This works well for most assets because criticality often *does* correlate with how connected or heavily used something is.

### When to Use Manual Override

Auto-calculated criticality can miss business context that usage data alone doesn't capture. Switch to **Manual** when:

* **Regulatory or compliance tables** — rarely queried, but legally mandated and high-risk if incorrect (e.g., financial reporting source tables).
* **Low-traffic but high-stakes assets** — used only during specific events like month-end close, audits, or annual reporting.
* **New or recently launched assets** — not enough usage history yet for Auto to reflect true importance.
* **Strategic initiatives** — assets supporting a new business priority that hasn't generated significant query volume yet.

### Why This Matters

Criticality classification drives:

* **Alert prioritization** — critical assets typically trigger higher-severity alerts on quality issues.
* **Review urgency** — issues on critical assets get flagged for faster resolution.
* **Governance focus** — stewards and owners are encouraged to prioritize documentation and monitoring for critical assets first.

Manual override ensures these downstream behaviors reflect actual business risk — not just system-observed activity.