The 2026 Gartner Magic Quadrant for MDM is back after four years. Here is what it does not score.

Gartner retired the Magic Quadrant for master data management, and then in April 2026 published one anyway. It is the first in four years, it runs to roughly fifty pages, and it scores twenty vendors. Within about a week, every vendor named in it had a press release up and a gated reprint behind a form.

The report is good. Gartner analysts talk to reference customers, and the vendor cautions in these documents are usually the most honest paragraphs published about enterprise software all year. My problem is not with the quadrant. It is that the people I talk to most keep being handed this chart as though it were a shortlist. They have just found out Master Data Services is gone in SQL Server 2025, they need a plan by Q1, and the chart is going to send half of them down a road they cannot afford. It is not a shortlist. The reason is written into the first page.

Who is on the chart

Gartner does not publish the full grid outside the paid report and the vendor reprints, so the public picture is assembled from what each vendor announced about itself. Five have claimed Leader placement:

Pimcore has announced placement as a Niche Player. The other fourteen vendors are in the report without having made a public announcement about where they landed, which tells you something on its own about how these documents get used.

One caveat on reading vendor announcements as data. Every one of them is written by the vendor, quotes the part of the report that flatters it, and legally cannot quote the cautions. Five companies all describing themselves as Leaders is consistent with the chart and tells you nothing about the ordering within it.

The market definition is the whole story

Gartner scopes this report to solutions that master, cleanse, enrich, link and synchronise multiple data domains, with governance and stewardship across the estate. Two words in that sentence do more work than the entire evaluation that follows.

If you master one domain, the report never evaluates you at all. That is different from ranking badly, and the difference matters: nobody read your documentation and found it thin. Gartner is describing the enterprise MDM market, the enterprise MDM market really is multi-domain, and a market definition doing its job is not a scandal. The failure happens downstream, when somebody hands the chart to a buyer as a list of every tool worth considering.

There is a second filter underneath it. Inclusion criteria for these reports set a revenue floor and a minimum count of reference customers. Gartner does not publish the 2026 numbers in the material available outside the report, and I am not going to quote the 2017 edition’s figure as though it were current. What is observable without the document: every vendor on the grid is an established multi-domain platform business, and no vendor selling a focused single-domain tool appears anywhere on it. That pattern is the criteria showing through.

So the quadrant answers one question well: among vendors that already sell enterprise multi-domain MDM, which are strongest? If that is your question, buy the report. If your question is which single domain to fix first, the report is not scoped to it.

Four things the axes do not score

Completeness of Vision and Ability to Execute are both about the vendor. Neither is about your implementation. These four decide most mid-market outcomes and appear on neither axis.

Number two is the one that catches people, because it inverts. A deep partner network is a real strength for a company running a multi-year programme across five domains, and Gartner is right to score it. For a team of two with a Q1 deadline it is the exact thing that makes the platform unbuyable. The same fact, read from two budgets, comes out with opposite signs.

Number three matters more than it used to. Microsoft removed MDS from SQL Server 2025 and did not replace it, which pushed a large group of on-premises teams into the market at once. Many of them cannot move master data to a hosted platform for reasons that have nothing to do with preference. For those teams, most of the grid is unavailable regardless of position.

What it is genuinely good for

Plenty, and I would still download the reprint.

The cautions are the best part of any Magic Quadrant and this one is no exception. Every vendor profile carries a section on where that product struggles, written by analysts who spoke to people running it in production rather than to the sales team. Vendors cannot reproduce those paragraphs in their own marketing, which is why they are worth reading. If you have already narrowed to one of the five, skip everything else in the document and read that vendor’s cautions twice.

It doubles as a map of who is still standing after four quiet years, one acquisition folding Informatica into Salesforce, and a lot of repositioning around AI. If you last looked at this market in 2021, that alone is useful.

Then there is procurement cover, which nobody puts in a business case but everybody uses. If your committee wants an external reference before it signs anything, this is the document that settles the room. Wanting that is not a failure of nerve. It is just how buying works in an organisation big enough to have a committee.

We are not on it

Primentra is not in the 2026 Magic Quadrant, and it would be odd if it were. We are a small vendor. We sell a tool that runs on your own SQL Server, that most customers point at one domain to begin with, and that a two-person team installs in about a day. Every one of those sentences is a reason we fall outside the market this report defines, and none of them is something I would change to get on a chart.

The honest version of the trade: we have no probabilistic matching engine, no multi-domain federation across a dozen source systems, and no partner network. If you are reconciling two million customer records from six acquisitions with no shared key, buy from the quadrant and budget for the implementer. Those vendors are on it because they are good at that, and we are not pretending otherwise. Our 2026 alternatives comparison puts the wider field side by side with pricing attached, and the mid-market argument is the longer version of why the analyst view and the two-person view diverge.

What I would push back on is the inference. A quadrant is a picture of one market segment at one moment, drawn to a definition that excluded most of the software people actually run. Read it for the cautions. Then go and answer the four questions it does not score, because those are the ones that will decide whether your project works, and there is nobody who can answer them on your behalf.

Common questions