Core Concepts

The building blocks of data observability

Data observability answers a simple question: Can I trust this data?

When a dashboard shows unexpected numbers, you need to know if it’s a real business trend or a broken pipeline. When an executive asks about yesterday’s revenue, you need confidence that the data is fresh and complete. Data observability gives you that confidence.

The Problem

Data pipelines fail silently. Unlike application errors that crash loudly, data problems often go unnoticed until someone makes a bad decision:

Scenario	What Happened	The Cost
Marketing spends $50K on wrong audience	Pipeline dropped demographic column	Wasted ad spend, wrong targeting
CEO quotes wrong revenue in earnings call	ETL job failed, dashboard showed stale data	Stock price impact, credibility loss
Product team ships feature to wrong segment	Upstream table had schema change	Development time wasted, wrong launch

Data observability prevents these scenarios by monitoring your data like you monitor your applications.

The Building Blocks

AnomalyArmor monitors data through these interconnected concepts:

Assets

Tables, views, and other data objects that AnomalyArmor monitors

Discovery

How AnomalyArmor finds and catalogs your data assets

Schema Changes

Detecting and tracking structural changes to your data

Freshness

Monitoring when your data was last updated

Metrics

Tracking statistical properties and detecting anomalies

Alerts

How you get notified when issues occur

Report Badges

Embedded status indicators for dashboards and docs

Intelligence

AI-powered search and documentation

Tagging

Classifying and organizing your assets

How They Work Together

Discovery scans your databases and catalogs Assets (tables, views, columns). Once cataloged, AnomalyArmor monitors each asset for:

• Schema changes: Columns added, removed, or type-changed
• Freshness violations: Data not updated within your SLA
• Metric anomalies: Unexpected changes in row counts, null rates, or distributions

When issues occur, Alerts notify your team through Slack, PagerDuty, email, or webhooks.

Report Badges embed this status directly in your dashboards and documentation, so consumers always know if data is trustworthy.

Intelligence helps you explore your catalog with natural language, and Tagging organizes assets for compliance and governance.

Quick Reference

Concept	Question It Answers	Example Alert
Asset	What data do I have?	(Cataloging, no alert)
Discovery	What changed since last scan?	”New table detected: staging.orders_v2”
Schema Change	Did the structure change?	”Column removed: orders.shipping_status”
Freshness	Is data up to date?	“orders table is 4 hours stale”
Metric	Is data quality normal?	”Row count dropped 60% from yesterday”
Alert	Who needs to know?	(Routes to Slack, PagerDuty, etc.)
Report Badge	Can consumers trust this?	(Visual indicator on dashboards)
Intelligence	Where is X data?	(AI-powered search result)
Tagging	What category is this?	(Classification: PII, production, etc.)

Next Steps

Connect Your Database

Start monitoring in under 15 minutes

Explore Assets

Understand the foundation of data observability