Search for data governance tools and you will find eight lists on the first page. Most of them were published by a company that appears in its own list, usually near the top. That is not a scandal, it is content marketing, but it does mean the lists are organised around who paid for the page rather than around what you are trying to fix.
We sell one of these tools, so treat this page the same way. What follows is the split that decides the purchase, and the point at which each option stops being the right answer, including ours.
The category is being renamed underneath you
We pulled twelve months of search volume against the twelve before it in August 2026. The numbers are worth knowing before you brief anyone internally:
| What people search for | Year before | Last year | Change |
|---|---|---|---|
| data governance software | 2,160 | 25,060 | +1060% |
| master data management software | 3,080 | 12,300 | +299% |
| mdm tools | 6,900 | 13,620 | +97% |
| product information management | 9,590 | 9,720 | +1% |
| master data management | 56,800 | 26,900 | -53% |
United States, English, annual totals from Google Ads keyword data via DataForSEO, pulled 31 August 2026.
The work did not change. The word did. Anyone whose internal business case still says master data management is describing the same project in the vocabulary of two years ago, and will find the budget conversation harder than it needs to be.
The split that decides everything
Every tool in this category is on one side of a line, whatever the marketing says.
Catalogue tools read. They connect to your systems, inventory the tables, infer the lineage, and give you a searchable map with owners and definitions attached. Collibra, Alation, OpenMetadata, DataHub. Nobody edits a customer address in a catalogue, and that is by design.
Master data tools write. They hold the shared record itself, refuse values that break a rule, route the risky changes to an approver, and keep a history of who changed what. Semarchy, Profisee, SAP MDG, Primentra.
Buying the wrong side is the expensive mistake in this category, and it is easy to make, because both halves answer the question “do we have data governance?” with a yes. Only one of them stops the next bad supplier record.
The list, with the stopping point included
| Tool | Side of the line | Who it suits | Where it stops |
|---|---|---|---|
| Collibra | Catalogue + governance | Large enterprises with a governance team already in place | Priced and scoped for a programme, not a project |
| Alation | Catalogue | Analytics teams that need to find and trust existing data | Documents data; does not decide what a record says |
| Informatica | Full suite | Estates already running Informatica pipelines | Acquired by Salesforce in November 2025; roadmap is in motion |
| Semarchy | Master data + governance | Mid to large estates wanting both in one platform | Real implementation effort; not a two-week install |
| Profisee | Master data | Microsoft-centric estates moving off MDS | Quote-based; sized for the upper mid-market |
| OpenMetadata / DataHub | Catalogue (open source) | Teams with engineers who will own it | You are the support contract |
| Primentra | Master data + governance | SQL Server estates that lost MDS and want rules, approval and history back | Not a catalogue. It holds records, it does not map your lake |
One entry deserves a note. Salesforce completed its acquisition of Informatica in November 2025. Nobody outside the company knows yet what that means for the on-premises products, and a five-year platform decision taken during someone else's integration is a decision made with less information than usual.
Four questions that shorten the evaluation
- What breaks today? Cannot find the data is a catalogue problem. Three systems disagree is a master data problem.
- Can I get a working instance this week? If the answer involves a scoping call before a trial, add a quarter to your plan.
- Does the rule run on the API too? Ask for a demonstration, not a yes. A rule enforced only in the screen is a rule with four ways around it.
- What does year three cost? Per-connector and per-user pricing grows with adoption, which means success is billed.
Where we fit, and where we do not
Primentra is on the write side. It holds the records, runs the rules on every route in, routes changes to approvers, and keeps history that a non-technical person can read. It runs on your own SQL Server, which matters if your data is not allowed to leave the building.
It is not a catalogue. It will not crawl your data lake, infer lineage across forty systems, or tell an analyst which report depends on which column. If that is the problem, buy Alation or stand up OpenMetadata and be happy. We would rather say so on this page than three weeks into a pilot.
Common questions
Do we need a catalogue or a master data tool?
Ask what breaks. If people cannot find data or do not know what a column means, that is a catalogue problem. If three systems hold three different addresses for one supplier, a catalogue will document all three beautifully and change none of them. That is a master data problem.
Can one tool do both?
Several claim to, and the honest answer is that they do one well and the other adequately. Buying the combined suite usually means paying enterprise prices for the half you need and inheriting a rollout for the half you do not.
Why does nobody publish pricing?
Because the price depends on how many connectors, users and domains you end up with, and because a quote is the start of a negotiation. It is a legitimate model for a large deployment. It is also why a two-week evaluation turns into a two-quarter one.
Is open source a real option?
For cataloguing, yes: OpenMetadata and DataHub are used seriously in production. For master data, the open source options are thin, because the hard part is not storage. It is rules, approval and history, and that is a lot of software nobody has given away.
Related reading: master data management tools for SQL Server estates, data governance vs master data management, and MDM vs data catalog.
If you landed on the write side of the line
You can have a working instance this afternoon rather than after a scoping call. Primentra installs on your own SQL Server in about a day. €7,500 per year flat, published, with a 60-day trial and no per-connector maths.