Can I trust the numbers right now — what broke, who owns it, and what's downstream?
Synced 2m agoMonte Carlo · dbt freshness · Atlan lineage
96/ 100
Data Health
Composite Trust
▲ 2 over 7 days
Fresh
Volume
Schema
Quality
Tables Monitored+coverage
2,847
78% certified ▲ 3%
Active Incidents · MTTR
2 both Low
MTTR 1h 47m▼ 22%
Freshness SLA
99.1%
met · 3 volume anomalies (7d)
Test Pass Rate
98.6%
12,904 / 13,082 passed
Incident Feed
2 active · 4 resolved 24h
Column-Level Lineage
quality badges propagate downstream →
Blast radius
14 downstream assets · 3 dashboards in path of stg_payments
Healthy & certifiedWatch · anomalyBI dashboard
Reliability Over Time
30 days · band 89–97
91 · 30d ago96 today · no breach 9d
Certification & Glossary
data products
Auto-certification recommenderactive_sessions now stable enough to certify — 31 days freshness met, 0 failed tests. Recommend certify.
Anomaly Timeline
expected band vs actual row counts
orders_daily · row volume · last 12h
expected 1.18M–1.24M
BREACH 02:40actual dipped to 0.97M rows — recovered 03:05, auto-clustered to upstream late-arriving partition.
Root-cause overlaySchema change in stg_payments.currency_code 41 min ago → distribution shift in 2 downstream models. Draft incident summary & suggested owner @r.lopez ready.
Monitor Coverage by Domain
monitors · freshness · owner
Domain
Monitors
Fresh
Owner
Finance
412
99%
@finance-eng
Growth
388
98%
@growth-data
Product
521
99%
@product-data
Payments
274
96%
@payments-eng
Marketing
196
99%
@mktg-analytics
How to use this
Reading the reliability board
What an Analytics Engineering / Data Reliability lead does with this view, top to bottom.
1Start at the Data Health Score. It rolls freshness, volume, schema and quality into one 0–100 number. A drop here is your "trust is at risk" alarm before any single incident looks scary — drill the score components to see which axis moved.
2Triage the Incident Feed left-to-right. Cards are ordered by severity then time-to-detect. Low TTD on a high-sev card is good (you caught it fast); a stale card with no owner avatar is the one to escalate.
3Trace the lineage graph to scope impact. Quality badges flow downstream along the bezier edges — a red dot on stg_payments tints everything to its right. Follow the curves to see exactly which dashboards inherit the problem.
4Let the AI do the clustering. Root-cause overlay ties an anomaly to a likely commit and drafts the incident note; blast-radius estimator counts downstream assets and affected viewers so you notify the right people first, not everyone.
5Certify what's earned it. The auto-certification recommender promotes only tables with a clean freshness + test history — certification is a contract with consumers, not a vanity badge.
For your own org: if a board exec pulled a number right now and it was wrong, how many hops back would you have to walk by hand to find the cause — and would you even know which dashboards were affected?
In context
Sample feed
Data reliability signal
Cross-stack reliability indicators a live integration could surface beside your own.
dbtSource freshness — 41 of 42 sources within SLA; shopify_orders 12m past warn▼ 1
SnowflakeQuery failure rate — 0.18% over 24h, well below 0.5% guardrail▲ stable