Say your Datadog dashboards are green. CPU is fine, the API’s p99 is fine, the warehouse isn’t throwing errors. The revenue table loaded yesterday’s file for the third day running, and nothing in Datadog knows that’s wrong. It’s not built to.
That’s why the two run side by side without much fighting.
They answer different questions.
Two kinds of monitoring, two questions
| Application and infrastructure monitoring | Data observability | |
|---|---|---|
| Examples | Datadog APM, New Relic, cloud provider monitoring, Splunk logs | Freshness, volume, and schema monitors, data tests, pipeline run tracking |
| Watches | Servers, services, latency, error rates, logs | Tables, columns, test results, job order and timing |
| Answers | Is the system up and fast? | Did the system produce the right data, on time? |
| Misses | A healthy job loading the wrong file | A full disk, unless a job fails because of it |
| Who fixes it | Platform or SRE team | Data engineering |
The DataOps Observability technical overview puts it bluntly: application monitoring checks like disk and CPU are often lagging indicators of problems with your data. IT observability tools are valuable and incomplete, because data problems span the raw data, the integrated data, and the code that drives the tools.
Make the alerts land in the same place
Running two systems is fine. Running two places to look is what fails. Point both at the channels your team already reads.
TestGen added Slack, Microsoft Teams, and webhook notification destinations in its October 14, 2026 release. Set one up in Project Settings and any notification rule can use it. DataOps Observability rules send email or a web request, and the web request is how its findings reach Slack, Jira, or Teams.
Going the other direction is less automatic. DataKitchen’s 2022 Taming the Chaos series argued that data observability has to accept logs and events from IT monitoring like Datadog, Splunk, and Azure Monitor, because pipelines run on infrastructure. Observability’s pre-built agents cover data tools (Airflow, dbt Core, Databricks, Azure Data Factory, Fivetran, Power BI and others). There’s no Datadog agent on that list. Infrastructure events come in through the Event Ingestion API.
The bundle question
If you already pay Datadog, the obvious move is Datadog’s own data observability. Metaplane went to Datadog, and the product now sits inside Datadog’s suite. The procurement path is short and the alerts are already in one place. Those are real advantages.
Check the pricing model before you assume the bundle is cheaper. DataKitchen’s Metaplane (Datadog) comparison estimates $60,000 to $150,000 a year for 10 users and 1,000 tables, per table or by consumption (DataKitchen’s estimate, reviewed June 2026). DataKitchen’s March 2026 post on $1 billion in data observability venture money cites a 2022 report that Datadog charged Coinbase $65 million for one year of IT observability. That’s an extreme case, and it’s also what land-and-expand pricing looks like at the far end.
The same post makes the other point: when a platform buys a point tool, the tool’s independent roadmap usually ends. If you were a Metaplane customer, you’re a Datadog customer now.
DataKitchen’s answer is its own stack, so read this as a vendor’s view: TestGen Enterprise lists at $100 per user and $100 per connection per month with no per-table fee, and the open-source editions of TestGen and Observability are Apache 2.0. Either way, keep Datadog for what it’s good at.