Data catalog pricing is difficult to compare because vendors meter different things and many enterprise products do not publish a complete price list. A low headline fee can also be outweighed by connector work, infrastructure or stewardship.
The defensible comparison is a three-year total cost of ownership built from your deployment assumptions.
Common commercial models
Vendors may charge by one or more of the following:
- named, active or role-based users;
- catalogued or governed assets;
- metadata storage or processing capacity;
- connectors or source types;
- environments such as development and production;
- optional modules for lineage, quality, privacy or AI;
- API usage or automation volume; and
- support or service tier.
Do not assume a familiar label means the same meter. “Asset” could refer to a source system, table, column, report or governed data product depending on the product and edition.
Microsoft’s current Purview governance model, for example, distinguishes governed assets and processing units; its billing FAQ explains daily asset counting and which Data Map activities are treated separately. That model should not be generalised to other vendors.
Licence cost questions
Ask each vendor to price the same written scenario and disclose:
- The precise unit being counted.
- How usage is measured and reported.
- Included minimums, allowances and overage rates.
- Which demonstrated capabilities require another edition or module.
- Separate charges for non-production environments.
- Expected renewal increases and price protections.
- Costs for support, training, implementation and migration.
- Contract assumptions that would change the quote.
Request a sensitivity table for growth in assets, users and sources. A quote that works only at today’s volume is not a three-year estimate.
Open source does not mean zero cost
An open-source catalog can remove the proprietary licence fee while increasing infrastructure and engineering responsibility. Include:
- compute, databases, search infrastructure and storage;
- Kubernetes or other platform operations;
- upgrades, backups, monitoring and disaster recovery;
- connector configuration and break/fix work;
- authentication and authorisation integration;
- security review and vulnerability response;
- support contracts if required; and
- engineering opportunity cost.
Products such as DataHub and OpenMetadata publish open-source repositories under Apache 2.0, but each also sits within a wider commercial ecosystem. Verify the licence of the exact repository, distribution and feature you plan to deploy rather than relying on a category label.
Implementation cost
Initial implementation commonly includes architecture, security approval, source onboarding, metadata modelling, taxonomy work, migration, training and change management.
Estimate effort by role and phase. Separate vendor or consultancy fees from internal time; internal work is still a cost even when it does not create a purchase order.
Connector depth affects this number substantially. A native connector that captures the required lineage may avoid custom development, while a connector that only extracts schema can leave the harder work to your team.
Ongoing operating cost
The catalog needs an owner after launch. Budget for:
- platform administration and ingestion monitoring;
- source changes and credential rotation;
- user support and training;
- stewardship and glossary decisions;
- quality and ownership reviews;
- upgrades and release assessment;
- usage measurement; and
- periodic retirement of stale assets.
Stewardship is often the largest hidden cost. It should also be counted as value-producing work: the point is not to minimise stewardship hours blindly, but to focus them on decisions that improve discovery and trust.
Build a three-year model
Create rows for licence, implementation, infrastructure, external services and internal roles. State quantities and rates separately so assumptions can be challenged.
Model at least three cases:
- Base: expected adoption and estate growth.
- High growth: faster increase in governed assets, users and sources.
- Operational stress: additional connector maintenance, migration or support.
Discounts should be shown explicitly rather than baked into unit rates. Record whether they expire at renewal.
Compare cost with the value case
Avoid fabricating a precise productivity return from optimistic time savings. Tie value to the chosen workflow and measure a baseline during the pilot.
Examples include reduced analyst search time, faster impact analysis, fewer duplicated assets, shorter audit evidence preparation and lower incident recovery time. Document who benefits and whether time saved can realistically be redeployed.
Red flags in a quote
- Undefined units such as “catalog capacity.”
- Critical modules omitted from the proposed edition.
- Connector charges shown without required infrastructure or services.
- A pilot price presented as the production run rate.
- No export or termination path for accumulated metadata.
- Discounts that conceal a large renewal step-up.
- Professional services required to perform routine administration.
Pricing changes. Use current vendor quotes and official pricing material for procurement rather than figures repeated in undated comparison articles.
Next, apply these cost questions within the software evaluation process or compare the operational trade-offs in open-source versus commercial catalogs.