Which data quality tools generate tests automatically?

DataKitchen's modern data quality roundup lists six that do it without qualification: DataKitchen TestGen, Anomalo, FirstEigen, Telmai, DQLabs, and Qualytics. Many others, including Monte Carlo, Bigeye, and Soda, rate partial, meaning suggestions you still review and author. DataKitchen makes TestGen, which generates 34 test types from a profile of 55 column characteristics.

How much do data quality testing tools cost in 2026?

At 10 users and 1,000 tables, most enterprise tiers land between $60,000 and $250,000 a year, by DataKitchen's estimates reviewed June 2026. Few vendors publish list prices, so treat those as ranges. TestGen Enterprise lists at $100 per user and $100 per connection per month, $12,000 to $36,000 a year at that size.

Are auto-generated data quality tests enough?

No. Generated tests learn what your data already looks like, so they pass on data that is consistently wrong. In DataKitchen's 2026 NYC taxi test, 62 generated tests passed on a 300,000-row sample whose minimum fare was negative $367. On the full 2,964,624 rows, 11 of 69 tests failed, including 7 business rules an AI wrote from the profile.

Which data quality tools test code changes before they reach production?

Recce is built for that job: it compares development and production data on a dbt pull request before merge, and it is open source. Datafold built its name on the same data diffs, though since March 2026 it leads with AI-powered migrations. dbt build stops downstream models when a test fails. TestGen's REST API can trigger profiling and test runs from a CI/CD pipeline.

What should I ask a data quality vendor in a demo?

Ask how tests get created, since hand-authoring is what caps coverage. Ask where tests execute, because running inside your warehouse keeps rows on your network. Ask how the bill changes when your table count doubles. And ask how many of its MCP tools can write, not just read; TestGen's open-source MCP server has 96 tools.

What are the best data quality testing tools in 2026?

The best data quality testing tool writes most of your tests, catches breakage before a user does, and runs inside your existing workflow. DataKitchen's comparison pages list six that generate tests without qualification, including DataKitchen's own TestGen, Anomalo, and Qualytics. At 10 users and 1,000 tables, most enterprise tools cost $60,000 to $250,000 a year, by DataKitchen's June 2026 estimates.

Every vendor demo shows a red test turning green. The question a demo never answers is who wrote the test, and how many more of them you’ll need.

DataKitchen makes DataOps TestGen and scores it in the same tables as everyone else. Check every row on its linked comparison page.

What to evaluate

Criterion	What to ask	Why it matters
How tests get created	Generated from a profile, suggested for review, or hand-written in YAML, SQL, or Python?	The coverage guide sets a floor of two tests per table plus tests on every column. Hand-written coverage stalls long before that
Business rules	Can a person add a rule without writing code?	Generated tests miss what the data never showed them
Catching breakage in production	Freshness, volume, and schema monitors on every table, or only the ones you priced?	Per-table pricing leaves tables unwatched
Catching breakage before merge	Can it run in CI and fail the build?	A code change can move a number with no error anywhere
Where tests run	In your database, or by copying rows to the vendor?	In-database execution keeps data on your network
Workflow fit	REST API, Git-friendly definitions, an MCP server that can write	Tools your engineers can’t script get ignored
Price model	Per user, per table, per credit?	The bill should not grow every time you cover another table

How the main tools compare

Ratings and prices come from DataKitchen’s comparison pages, reviewed June 2026, sized for 10 users and 1,000 tables. Prices are DataKitchen’s estimates, not quotes.

Tool	Auto-generated tests	Freshness, volume, schema	MCP server	Deployment	Estimated annual price
DataKitchen TestGen	Yes	Yes	Yes, 96 tools in open source	Self-host; managed cloud for Enterprise	$0 open source; $12,000 to $36,000 Enterprise
Anomalo	Yes	Yes	No	SaaS, self-host option	$100,000 to $200,000
Qualytics	Yes	Yes	Yes	SaaS, self-host	$60,000 to $150,000
DQLabs	Yes	Yes	No	SaaS	$60,000 to $150,000
Telmai	Yes	Yes	No	SaaS, BYOC	$80,000 to $200,000, volume-dependent
FirstEigen DataBuck	Yes	Yes	No	SaaS, self-host	$60,000 to $150,000
Monte Carlo	Partial	Yes	Partial (beta)	SaaS	$120,000 to $250,000
Bigeye	Partial	Yes	Partial (beta)	SaaS	$80,000 to $180,000
Soda Enterprise	Partial	Yes	No	SaaS	$100,000 to $200,000
Validio	Partial	Yes	Yes	SaaS, self-host option	$80,000 to $180,000
Datafold	Partial	Partial	Yes	SaaS, self-host option	$100,000 to $180,000
Recce	No	Partial, PR diffs only	Yes	Self-host, SaaS	$0 open source; Team tier $250 a month
AWS Glue Data Quality	Partial	Partial	Yes	AWS managed	$5,000 to $30,000, usage-based

TestGen doesn’t lead every column. Its column-level lineage rates partial, where Monte Carlo, Bigeye, and Qualytics rate yes. Open-source TestGen covers one user, one database connection, and one project. For free tools only, see the open-source answer.

Catching breakage early takes two kinds of check

Production breakage shows up as a table that stopped loading, lost half its rows, or dropped a column. Table monitors catch those with no rules to write.

Deployment breakage is quieter. A transform that compiles fine starts returning a different number. Only a regression test on the same input catches it, and that test has to run before the merge. Recce is built around that comparison. Datafold was too, until it moved its lead product to AI-powered migrations in March 2026. The diff product still exists. dbt’s dbt build stops downstream models when a test fails. TestGen’s REST API can trigger profiling and test runs from a CI/CD pipeline, per its API documentation linked on each comparison page. For the reasoning behind both, see shift-left testing.

Generated tests are a floor

The NYC taxi test is the clearest example of what generation does and doesn’t do. TestGen generated 62 tests from a 300,000-row sample, and all of them passed, because the minimum-value test set its floor at the lowest fare present: negative $367. On the full 2,964,624 rows, 11 of 69 tests failed. Four were generated minimum-value tests. Seven were business rules an AI wrote after reading the profile.

So the best tool writes the bulk. A person writes the rest.

DataKitchen’s 80% post puts the generated share at about 80% of tests. The rest is domain logic.

On price

DataKitchen’s estimates for the enterprise vendors on its comparison pages averaged $172,000 a year as of April 2026, per a post on how crowded the vendor field is. Few of those vendors publish list prices. The category has taken more than a billion dollars of venture money, which is part of why prices look the way they do. For a small-team budget, see the cost answer.

Sources

Source	What it covers
The 2026 commercial data quality and observability software map	More than 50 commercial vendors in four categories, with 2026 ownership changes and Datafold’s move to migrations
Best modern data quality tools	The scored roundup the ratings come from

What are the best data quality testing tools in 2026?

Key points