Where this data comes from
sku2name parses what Microsoft publishes and presents it in a searchable form. It maintains no catalog of its own.
Two sources, because neither is complete
Microsoft publishes the same mapping twice, and the two files do not agree. Rather than pick one and silently lose records, sku2name reads both and merges them by GUID.
- The markdown reference in the entra-docs repository. It lives in public source control, so a change is attributable to a commit.
- The CSV export linked from that page. It has one row per SKU and service plan pair with a dedicated service plan ID column, so it is far cleaner to parse.
In the current sync, 0 SKUs appear only in the CSV, and 30 SKU-to-plan relationships appear only in the markdown. Using either file alone would silently drop real records.
Current sync
- Data last changed
- 30 Sep 2026
- Microsoft last updated
- 19 Aug 2026
- Document ms.date
- 2026-08-18
- Upstream commit
- 872fba653898
- SKUs
- 620
- Service plans
- 797
- SKU to plan relationships
- 6,001
- Dataset hash
- 91b5d71fbe40ab0a
How a display name is chosen
257 service plans appear in Microsoft's data under more than one name. sku2name picks one canonically, using a fixed rule so the same input always produces the same output:
- Set aside a name whose brackets do not close:
Microsoft 365 Business Voice (UKloses toMicrosoft 365 Business Voice (UK). - Set aside a name that is another spelling with characters missing from inside a word:
Dynamics 365 P1 Trialoses toDynamics 365 P1 Trial. - Then prefer a name whose shape matches its kind:
UNDERSCORE_CAPSfor technical names, prose for display names. - Then prefer the current brand name over one Microsoft has renamed, so
Azure Active Directorygives way toMicrosoft Entra IDandFlowtoPower Automate. - Then prefer a name that both Microsoft files agree on.
- Then prefer the more frequent spelling.
- Then prefer a display name that differs from the technical name.
- Finally, sort alphabetically, so the result is never ambiguous.
The first two steps exist because Microsoft's CSV carries truncated names its markdown does
not, so the complete spelling is usually sitting in the other file. Only a damaged word
counts. A name that adds or drops a whole word is a rename rather than a defect, which is
why SharePoint (Plan 2) is not replaced by SharePoint Online (Plan 2).
The rename step is a fixed list of Microsoft's own rebrands, because a retired name can sit in either file depending on which was regenerated last. It never invents a name: a spelling is only set aside when Microsoft has published the newer one for that same identifier somewhere. Where they have not, the older name stands.
Every rejected spelling is kept, listed on the plan's page, and remains searchable. Where Microsoft provides no display name distinct from the technical name, sku2name says so rather than inventing one. That is the case for 12 plans.
Known caveats
- Some technical names in Microsoft's data contain stray spaces or are ALL CAPS. These are shown as published, with obvious whitespace defects repaired.
- The CSV's own last-modified date is unreliable: it has read older than the markdown while carrying newer SKUs. sku2name ignores it and compares content instead.
- A handful of entries in the conflict tables reference plans that are not in the main table. They are shown as plain text rather than broken links.
Raw data
The normalized dataset is available as JSON: skus.json, service-plans.json, source-meta.json.
Where the dates, former names and changelog come from
Microsoft publishes no changelog for the licensing reference, but the markdown half of it is maintained in a public repository. Replaying its commit history, and parsing every revision with the same code that builds this site, reconstructs one.
That is where First documented and Microsoft called this come from. A date means the identifier first appeared in Microsoft's published reference, which is not the same claim as when a product existed or went on sale. Where an identifier was already present in the earliest revision that can be parsed, the page says before that date rather than inventing an arrival.
Names Microsoft has since replaced stay searchable, so an identifier copied from an older agreement or export still finds the right page. Two kinds are deliberately not shown as former names: a retitling that only changes capitalisation or punctuation, and a spelling that is the current name with characters missing, which is an upstream typo rather than a name the product ever had. Both remain searchable.
The same reconstruction produces the changelog. Two kinds of edit are left out of it, both because they would drown the real changes:
- Retitlings that only change capitalisation or punctuation. In September 2022 Microsoft recased 90 SKU names in a single pass, and reporting those as renames would bury the handful that were real.
- An entry Microsoft removed and later restored. Because sku2name merges both of Microsoft's files, a SKU that drops out of one while the other still lists it never actually disappears from this site, so publishing a removal and a return would describe a change that did not happen. Only a removal that stuck is listed.
Attribution and license
Microsoft's licensing reference is published in the
MicrosoftDocs/entra-docs
repository under the
MIT License,
which covers its documentation as well as its code. Unusually for a Microsoft documentation
repository, both its LICENSE and LICENSE-CODE are MIT rather than
the content being Creative Commons. sku2name reads the same table from both formats
Microsoft publishes: the markdown, and the CSV the markdown links to as "a CSV version of
this table". That license requires the copyright
notice to travel with substantial portions of the work, so it ships with this dataset:
Copyright (c) Microsoft Corporation. Licensed under the MIT License.
Full text: NOTICE.txt, shipped alongside the JSON downloads.
What sku2name adds on top is its own: the reverse index from service plans to SKUs, the similarity between SKUs, the canonical name selection, and the URL slugs. None of that comes from Microsoft, and it is covered by sku2name's own MIT license.
The underlying facts, the identifiers and GUIDs themselves, are Microsoft's.