Features
Data observability tools that cover all five pillars
Freshness, volume, schema, anomalies, and lineage, monitored automatically on every table, with incident tracking and the integrations your stack already runs on.
The short answer
DataObservability monitors freshness, volume, schema, anomalies and lineage on every table of a Snowflake, BigQuery, Databricks or Redshift warehouse, on every plan. Higher plans add capacity and depth: more tables, longer history, dbt-native monitors and PagerDuty from Team, column-level lineage, multiple warehouses, SSO and an audit log from Scale. Plans start at 59 dollars a month billed yearly, with a 14-day trial and no credit card.
The 5 pillars
Five pillars, one connection
fresh 4m ago
Freshness
Know the moment a table stops updating on schedule. Freshness monitors set SLAs per table and alert you when a sync stalls, a job fails, or a load runs late, before a stale number reaches a dashboard.
Learn more+0.4% vs forecast
Volume
Row counts that spike or collapse are the first sign of a broken load. Volume monitors learn each table's normal range and flag drops and surges against a forecast band automatically.
Learn more0 changes 24h
Schema
A renamed column or a dropped field breaks everything downstream silently. Schema-change detection alerts you the instant a tables structure shifts and traces it to dependents.
Learn morenulls 0.1%
Anomalies
Distribution drift, null spikes, and out-of-range values caught by ML-tuned thresholds that learn each table and keep false alarms low, so the alerts you get are worth acting on.
Learn more247 tables mapped
Lineage
Lineage maps every table to the models, dashboards, and exports it feeds, so you see the blast radius of any incident at a glance and find root cause fast. Every plan maps table to table; Scale and Enterprise add column-level lineage.
Learn moreBy plan
Every feature and the plan that includes it
Prices are per month billed yearly. Monthly billing is 119, 359, 959 and 1,998 dollars.
Swipe to see all columns →
| Feature | Starter $59 | Team $179 | Scale $479 | Enterprise $999 |
|---|---|---|---|---|
| Freshness, volume, schema and anomaly monitors | Yes | Yes | Yes | Yes |
| Table-level lineage and blast radius | Yes | Yes | Yes | Yes |
| Column-level lineage | No | No | Yes | Yes |
| dbt-native auto-monitors from manifest.json | No | Yes | Yes | Yes |
| Alert channels | Slack, email | Slack, email, PagerDuty | Slack, email, PagerDuty | Slack, email, PagerDuty |
| Tables monitored | Up to 50 | Up to 250 | Up to 1,500 | Unlimited |
| Warehouse connections | 1 | 1 | Multiple | Multiple |
| Metric history | 7 days | 90 days | 365 days | 365 days |
| SSO and audit log | No | No | Yes | Yes |
| In-VPC deployment option | No | No | No | Yes |
Security
What the warehouse connection reads, and what it never touches
You create a read-only user. Monitoring works from the catalog, so your compute bill and your security review both stay small.
Catalog metadata
Table and view lists, column names and types, and the row-count estimates your warehouse already keeps. This drives schema-change and volume monitors without scanning a single table.
One value for freshness
Freshness reads the newest value in the timestamp column you pick, one aggregate per check. That tells you a load is late without pulling any of the rows behind it.
A capped sample for null rates
The null profile counts empty values per column across at most 5,000 sampled rows. Counts come back; the values themselves are not stored or shown anywhere.
Lineage from the catalog, even read-only
Dependencies come from view definitions and key constraints. Where the standard information schema hides objects a read-only user does not own, lineage reads the system catalog as well, so the graph stays complete without granting more access.
No writes, no row exports, no copies of your data. Enterprise adds an in-VPC deployment for teams whose policy keeps every byte inside their own network.
Resolve
Incident tracking from alert to resolution
Every break becomes one tracked incident, not ten pings. Assign an owner, see the root cause and downstream impact, and keep a history you can learn from.
orders freshness SLA breached
- Owner@you
- Root causeFivetran sync paused
- Impact3 consumers
- Time to alert0.8s
Integrations
Connects to the stack you already run
Monitor any of these from one console, with lineage spanning your warehouse through to BI.
Questions buyers ask
Data observability features FAQ
Which data observability plan includes column-level lineage?
Column-level lineage is on Scale and Enterprise. Starter and Team map lineage table to table, which is enough to see which models and dashboards sit downstream of a broken table. Column-level lineage adds which exact field feeds which, the detail you need when a renamed column or a type change breaks a single metric rather than a whole table.
Do all plans include the five pillars of data observability?
Yes. Freshness, volume, schema, anomaly and lineage monitoring run on every plan, including Starter at 59 dollars a month billed yearly. Plans differ in how many tables you can watch, how long metric history is kept, which alert channels are wired in, and whether dbt-native monitors, column-level lineage and SSO are switched on.
Can I monitor Snowflake and BigQuery from one account?
Yes, on Scale and Enterprise, which allow multiple warehouse connections under one workspace with a shared incident list and lineage graph. Starter and Team cover one warehouse connection each, which suits most teams that run a single Snowflake, BigQuery, Databricks or Redshift account.
Which plans send alerts to PagerDuty?
Team, Scale and Enterprise route alerts to PagerDuty as well as Slack and email, with severity set per monitor so only real breaks page the on-call engineer. Starter sends to Slack and email, which is how most small teams handle data incidents before an on-call rotation exists.
Does data observability read the rows in my tables?
Almost never. Monitors read catalog metadata such as table lists, column types and row-count estimates. Freshness reads one value, the newest timestamp in the column you choose. The null profile reads a capped sample of up to 5,000 rows to count empty values per column. Nothing is copied out and the warehouse user stays read-only.
All five pillars, live on day one
Connect your warehouse and watch monitors generate across every table. Transparent pricing, no credit card.