Data Management Tools: 8 Categories, 30 Platforms, and What Each One Costs

Data management tools are software that moves, stores, cleans, tracks, and shows a company's data. The term covers eight different jobs, and one product almost never does all of them. This guide sorts 30 tools into those eight categories with the billing unit, free tier, and open-source status each vendor publishes.

Prices below come from each vendor's own pricing page as of September 2026. Check the page before you buy, because plan cards move.

What are data management tools?

Data management tools are the software a company uses to deliver, control, protect, and improve its data across the data's whole life. That wording tracks the DAMA definition of data management, which lists governance, quality, security, architecture, metadata, and integration as the core knowledge areas.

In practice, a "data management tool" is one of eight different products. An ELT pipeline that copies Salesforce into BigQuery is one. So is the warehouse it copies into, the dbt project that models it, the catalog that documents it, and the dashboard tool that shows it. A buyer who searches for "data management tools" is usually looking for one or two of these, and the first job is figuring out which.

The market for all of this is big. Grand View Research expects the enterprise data management market to reach $221.58 billion by 2030, growing 12.4% a year from 2025. Bad data is expensive too. In IBM's 2025 survey, 43% of chief operations officers named data quality as their top data priority, and over a quarter of organizations estimated losses above $5 million a year from poor-quality data.

What are the 8 categories of data management tools?

The eight categories are data integration (ELT/ETL), data warehouse or lakehouse, transformation, orchestration, catalog and governance, data quality and observability, master data management (MDM), and business intelligence (BI). Each one solves a different problem, and the tools in one category rarely replace the tools in another.

Fivetran's own ranking page groups the space into five layers (integration, MDM, governance, warehousing, quality), and Estuary's covers four. Both skip transformation, orchestration, and BI. Those three are where a small data team spends most of its week, so they get their own sections here.

Which data integration and ELT tools should you compare?

The integration tools worth comparing are Erathos, Fivetran, Airbyte, Estuary, Weld, Kondado, Matillion, dlt, Bruin, and Nekt. They differ most in how they bill: total rows, changed rows, gigabytes plus connectors, compute credits, or a flat subscription.

ELT means extract, load, transform: the tool copies raw data into the warehouse first and you transform it there with SQL. ETL transforms before loading. Every tool in this table is ELT or supports it.

Erathos bills on total rows written to your warehouse, with no limit on the number of connectors. The Free plan covers up to 1M rows a month. Movement is $29/month for 2M rows with hourly updates. Pro is $250/month for 5M rows and adds change data capture (CDC), custom connectors, five-minute updates, and API access. The connector directory lists 139 sources and destinations, including BigQuery, Snowflake, Databricks, Redshift, Postgres, and ClickHouse.

Under the hood, Erathos extracts to a temporary cloud bucket, then loads into the warehouse with a bulk COPY command or the destination's equivalent, then deletes the temporary files. Incremental loads use a cursor column and need a primary key plus a date or datetime column on the source table. If a table has neither, it runs as a full refresh.

Fivetran charges by monthly active rows: distinct records that were inserted, updated, or deleted in a month. Initial syncs and unchanged rows during a re-sync are free. The Free plan includes 500,000 connection MAR. Standard syncs every 15 minutes with 700+ connectors; Enterprise syncs every minute and offers hybrid deployment. This model gets cheap when your data changes little and expensive when it churns.

Airbyte sells three plans. Standard bills on volume and syncs at most once an hour. Pro bills on Data Workers, dedicated compute units, and syncs every 15 minutes. Enterprise Flex runs on-premises or in a hybrid mode where Airbyte hosts the control plane and you host the data plane. The self-hosted version ships under the Elastic License 2.0, which is source-available and free to run but restricts offering it as a hosted service.

Estuary charges $0.50 per GB plus $100 a month per connector instance for the first six instances, then $50 each. The Developer tier is free up to 10 GB and two connectors. Estuary focuses on real-time streaming, so it fits when sub-minute latency is the requirement.

Weld, Kondado, and Nekt are the small-team options. Weld's Basic plan is $99/month billed annually for two connectors, one user, and daily syncs, with 5M MAR included. Kondado's entry plan is $19/month for four pipelines and 1M records; hitting the record limit pauses pipelines until the next month. Nekt bundles a managed warehouse with its pipelines and bills in credits, with a free plan of 40 credits, 3 sources, and 10 GB of storage.

dlt and Bruin are for teams who want code. dlt is a free Apache 2.0 Python library; dltHub is the paid managed runtime at $11,900/month billed annually. Bruin is an Apache 2.0 CLI that defines ingestion, SQL and Python transforms, and quality checks in YAML in one repo.

Which warehouse or lakehouse should store your analytical data?

Snowflake, BigQuery, and Databricks are the three warehouses most integration tools load into. All three bill compute separately from storage, so the cost driver is how many queries you run, not how much you store.

BigQuery is serverless: no instances or virtual machines to size, and on-demand queries can use up to 2,000 concurrent slots per project. Snowflake sells compute as credits and lets you buy on-demand month-to-month or pre-paid capacity at a discount. Databricks prices in DBUs, its normalized unit of processing power, and lists Data Warehousing at $0.22 per DBU. Storage, networking, and related costs come from your own cloud account.

Which one you pick usually follows your cloud. Erathos has destination connectors for BigQuery, Databricks, Snowflake, Redshift, Postgres, and ClickHouse, so the warehouse choice does not lock in the integration choice.

Which transformation and orchestration tools fit your team?

dbt is the default transformation tool, and Airflow or Dagster are the default orchestrators. Transformation turns raw loaded tables into clean modeled ones with SQL. Orchestration decides when jobs run and what runs after what fails.

dbt's paid plans limit "successful models built per month." Developer gets 3,000 and Starter gets 15,000. A model run counts each time a scheduled job builds it, so a 100-model project running hourly burns 72,000 builds a month and needs Enterprise (100,000). Running dbt Core on your own scheduler has no such limit.

Dagster+ credits count materialized assets and executed ops. Serverless compute adds $0.010 per minute; the Hybrid option, where you run the compute, has no compute charge. Airflow itself is Python workflows under Apache 2.0 with no vendor to pay, but you run the scheduler, database, and workers yourself.

An ELT tool with its own scheduler covers a lot of this ground before you need an orchestrator. Erathos runs each pipeline on a schedule from every 5 minutes to weekly, with retries on Pro. Orchestration becomes necessary when a dbt run has to wait for three different loads to finish, or when a failed job in one system must block a job in another.

Which catalog, governance, quality, and observability tools should you add?

Catalogs (DataHub, OpenMetadata, Atlan, Collibra) answer "what data do we have and where did it come from." Quality and observability tools (Great Expectations, Soda, Monte Carlo) answer "is this data right, and did it arrive." Both have strong open-source options, and both enterprise vendors sell by quote.

DataHub draws the line between self-hosting and managed plainly. Core is free to run, but you handle installation, configuration, upgrades, uptime, and troubleshooting with community support. Cloud is the same product hosted by DataHub with an SLA. That tradeoff applies to every open-source row in this article.

Great Expectations' GX Core is Apache 2.0 and runs as Python checks inside your orchestrator. Monte Carlo lists four tiers with no dollar amounts; Start covers up to 10 users, up to 1,000 monitors, and 10,000 API calls a day. Soda publishes a $0 Free plan and a $750/month Team plan, billed in Soda Processing Units.

A catalog earns its cost when people cannot find the right table, or when a regulator asks where a column came from. A quality tool earns it the first time a dashboard shows stale numbers for a week and nobody notices.

When do you need master data management and BI tools?

Master data management (MDM) is for when the same customer, product, or supplier exists in five systems with five different IDs and you need one trusted record. BI is for turning modeled tables into dashboards people open. Most companies need BI on day one and MDM much later, if ever.

Informatica's MDM is sold through its consumption model, where IPUs give access to all eligible cloud services and MDM is metered on records per domain. Reltio sells a cloud-native SaaS multidomain MDM with entity resolution and reference data management. Neither publishes a dollar price.

For BI, the pricing unit is almost always a seat. Metabase's open-source edition is free with unlimited users if you host it. Power BI Pro is $14 per user per month, and Microsoft notes the price varies by region. Looker separates a platform fee from user licenses and sends you to sales for the number.

How do pricing models change the cost of data management tools?

Six billing units show up across these 30 tools: total rows, changed rows (MAR), records, gigabytes plus connector instances, compute credits, and seats. The same workload can cost very different amounts under each one, so the billing unit matters as much as the list price.

Two examples show the gap. Erathos counts every row written to the warehouse, so a 5M-row table on a full refresh counts 5M rows per run. Fivetran counts only rows that changed and excludes the initial sync, so that same static table costs almost nothing after day one. Flip the workload to an events table that appends 5M new rows a day, and both count 5M per day.

The way to keep row-based costs down is incremental loading. Erathos's partial refresh uses a cursor column so each run writes only rows past the last cursor value. CDC on the Pro plan reads the database's change log instead. The cursor vs CDC guide covers when each one makes sense.

Do small teams need every category of data management tools?

A small team needs three of the eight categories to start: an integration tool, a warehouse, and a BI tool. Transformation comes next when raw tables stop being enough for dashboards. The other four are for specific problems that have not happened yet.

Fivetran's own list of eight tools describes them as specialized products assembled alongside each other, and that is the honest picture. There is no single "data management tool." There is a stack, and the stack grows one problem at a time.

The cheapest working stack today is a free integration tier, a warehouse free tier, and open-source BI. Erathos Free covers 1M rows a month, BigQuery gives 1 TiB of queries and 10 GiB of storage a month free, and self-hosted Metabase is free with unlimited users. That runs a small company's reporting for $0 in software. The pipeline tutorial walks through the integration piece.

Try Erathos free for 14 days

Every new Erathos account gets the Pro plan for 14 days: CDC, five-minute syncs, custom connectors, and 5M rows. Each new connection also gets its own two-week window to test the connector and see what it does to your row count. After the trial, the account moves to Free if you have not picked a plan.

Try Erathos free for 14 days