Ranks Two, Three and Four in Canada's Federal IT Vendor Table Are the Same Company

What the file actually contains

Every federal contract award above the disclosure threshold is published, quarterly, by the department that awarded it (Treasury Board of Canada Secretariat, 2026). That is the reason federal competitive analysis is the one market-share slide people trust: the underlying records are public, so the number feels audited.

Here is what a working pull looks like. It comes in three nested layers, which have to be kept apart.

Layer one — the contract slice. Across 104 organizations carrying at least one information-technology contract, the file holds 6,000 IT contract records worth CA$1.49 billion (Treasury Board of Canada Secretariat, 2026).

Layer two — the named suppliers. Each organization's top suppliers, up to six apiece, give 573 rows under 360 distinct vendor strings, worth CA$1.18 billion — 79.1% of the contract value (Treasury Board of Canada Secretariat, 2026). Every concentration figure below sits on this layer.

Layer three — the individual awards. Each organization's most recent IT contracts, again up to six, give 578 individual award records worth CA$931.15 million, each carrying a date, a value, a description of what was bought (Treasury Board of Canada Secretariat, 2026). Every deal-size and category figure below sits on this layer.

Before any of it: what was pulled is each organization's 250 most recent contracts, not a census of federal procurement, and 14 of the 104 organizations returned fewer because they have not awarded that many.

The spine. Proactive contract disclosure was built so a citizen can look up an award. It was not built to be added up, and it lacks the two things any market-share calculation needs — a vendor identity key and a common time base. Microsoft appears under twelve strings, holding ranks 2, 3 and 4 at once with entries differing only in capitalisation and a trailing full stop. A pull of the 250 most recent contracts covers five days at National Defence and 7,331 days at the Military Grievances External Review Committee. Repair both and the file stops being a ranking and becomes a market description — one saying the median federal IT award is CA$338,316, that 162 of 248 suppliers hold exactly one federal relationship, and that what most vendors call "the federal market" describes 24 companies.

Twelve Microsofts

Sorted by value, the top of the raw vendor table reads like this:

Rank	Vendor string	Value	Organizations
1	SAS Institute (Canada) Inc.	CA$203.23M	1
2	MICROSOFT CANADA INC.	CA$82.69M	17
3	Microsoft Canada Inc.	CA$53.35M	12
4	MICROSOFT CANADA INC	CA$52.39M	16
5	ORACLE CANADA ULC	CA$38.95M	8

(Treasury Board of Canada Secretariat, 2026)

Ranks two, three and four are one company. The three strings differ by case, plus whether someone typed a full stop after Inc. Nothing in the file says they are related, because the file has no field in which to say it.

Nine more Microsoft strings sit further down (ranks are within the 360-string table, not the whole file) — MICROSOFT CANADA at rank 32, MICROSOFT LICENSING GP at 52, Microsoft Canada at 76, Microsoft Corporation at 115, Microsoft Canada Inc at 148, MICROSOFT LICENSING, GP at 174, MICROSOFT CORPORATION at 176, Microsoft at 237, Microsoft (via Softchoice/SSC) at 293. Twelve strings, 59 rows, 54 organizations, CA$206.19 million.

That last string repays a pause. Someone typed the reseller into the vendor name — the channel becoming briefly visible inside a field with no room for it.

Three of six slots

The clearest illustration is small enough to check by hand. The Office of the Commissioner of Official Languages has six named top suppliers in this file. Three are Microsoft:

Vendor string	Value
MICROSOFT CORPORATION	CA$0.61M
Microsoft Canada Inc	CA$0.60M
Microsoft Canada Inc.	CA$0.25M

(Treasury Board of Canada Secretariat, 2026)

Read as published, this organization has three mid-sized software suppliers, half its top-six list competitive. Read correctly, one vendor holds CA$1.46 million and half the list is a single relationship. Three other organizations show the same pattern: Employment and Social Development Canada, Veterans Affairs Canada, the Office of the Commissioner for Federal Judicial Affairs each carry Microsoft twice under different strings. The file does not rank one organization's own suppliers correctly, which is a strong reason to doubt what it says about a market.

What merging the names does

Collapse case, punctuation, corporate suffixes, the Canada qualifier, and the 360 distinct strings become 248 suppliers — 31.1% fewer (Treasury Board of Canada Secretariat, 2026).

Concentration among named suppliers, before and after merging name variants

• Top 5 — raw strings36.5%
• Top 5 — merged45.5%
• Top 10 — raw strings47.9%
• Top 10 — merged56.4%
• Top 20 — raw strings60.7%
• Top 20 — merged69.4%

Merged, Microsoft reaches 51 organizations — nearly four times the breadth of Oracle, the next-widest named software supplier at 13 (Treasury Board of Canada Secretariat, 2026). Breadth is the number a competitor actually needs, and the one the raw file destroys most thoroughly, because a vendor present in many organizations under many spellings fragments in proportion to its own footprint.

Merging names is a judgment, not a fact

My merge rule is written above and you can disagree with it. Two Microsoft strings — MICROSOFT LICENSING GP and MICROSOFT LICENSING, GP, carrying CA$5.01 million between them — stay separate under it, because Microsoft Licensing, GP is a distinct legal entity from Microsoft Canada Inc. Someone analysing corporate-group exposure should merge them. Someone analysing contracting counterparties should not.

That is the honest position on every normalisation rule: it encodes a definition of the same supplier that depends on the question. Publish the rule with the number. A concentration figure whose merge rule is unstated is not reproducible, and the gap between 47.9% and 56.4% is entirely a product of that unstated rule (Treasury Board of Canada Secretariat, 2026).

The window is not a window

The naming problem is the one people notice. This is the one that matters more, and I have not seen it stated anywhere.

Pulling the 250 most recent contracts per organization sounds like a uniform sample. It is not a time window at all — it is a record count, and the period covered depends entirely on how often an organization contracts:

Days covered by each organization's 250 most recent IT contracts

• National Defence (25 Jun – 30 Jun 2026)5 days
• Public Services and Procurement Canada15 days
• Median across 104 organizations704 days
• Veterans Review and Appeal Board5,516 days
• Military Grievances External Review Committee7,331 days

Add those together and you are adding five days of National Defence to twenty years of a small tribunal, then calling the result a market. The error is neither small nor random: high-volume buyers are systematically under-counted, because their 250 records cover a week rather than a decade. National Defence — the largest federal IT buyer by planned spending — contributes less to the total than its position warrants, and no amount of name-merging fixes that.

The repair is to bound the pull by date rather than by record count, then accept the consequence: a fixed date window returns almost nothing for small organizations, so a federal supplier-share analysis is honest only within a size band.

What the repaired file is good for

Rankings are the least interesting thing in here. Once the names are merged and the sampling frame is stated, the same records answer three questions a vendor has to settle before writing a federal plan — and none of them is what is my market share.

What the government buys

Classify the 578 individual awards by what the department said it was buying, one category each, and federal IT resolves into two markets of almost identical size plus a rounding error:

578 recent federal IT awards, CA$931.15M, by what was bought

• Software licence and maintenance43.0% · CA$399.96M
• People — consultants and IT services42.3% · CA$393.58M
• Hardware and equipment7.9% · CA$73.42M
• Software purchase and development6.2% · CA$58.10M
• Other, telecom and unclassified0.7% · CA$6.09M

The two halves behave differently, and the difference runs opposite to intuition. A software licence or maintenance award has a median value of CA$312,466 across 180 records. A consultants-and-services award has a median of CA$849,619 across 198 records — 2.7 times larger (Treasury Board of Canada Secretariat, 2026). The individually large federal IT deals are people, not product. A software vendor assuming the reverse has mis-sized both the opportunity and the competitor set.

How big a federal IT deal is

This is the number I would put in front of a founder before anything else. Across the 578 awards:

	Value
25th percentile	CA$168,718
Median	CA$338,316
75th percentile	CA$1,229,463
90th percentile	CA$3,675,000
Largest single award	CA$103,424,997

The median is the midpoint of the 578 awards. The quartiles and the 90th percentile are the nearest actual award at each cut rather than an interpolation, so every figure in that column is a contract somebody signed.

(Treasury Board of Canada Secretariat, 2026)

72.8% of awards fall under CA$1 million. 38.8% fall under CA$250,000. The mean is CA$1.61 million, nearly five times the median (Treasury Board of Canada Secretariat, 2026) — what a distribution looks like when a handful of enterprise licence renewals sit on top of several hundred ordinary purchases.

That carries a hard consequence for go-to-market design. A sales motion built around a CA$2 million average contract value — dedicated capture manager, twelve-month pursuit, bid-and-proposal budget — is priced for the top decile and loses money across three-quarters of the market. Most federal IT selling has to survive on a mid-six-figure deal, which means either a light-touch motion or a channel that carries the cost.

Whether "the federal market" is one market at all

Merge the names, then count how many organizations each supplier reaches across that CA$1.18 billion. The distribution is the finding:

• 162 of 248 suppliers appear in exactly one organization, carrying 41.0% of the named value.
• 86 appear in two or more, carrying 59.0%.
• Only 24 suppliers reach five or more organizations. Those 24 carry 38.0% of the value.
• The median supplier's federal footprint is one organization (Treasury Board of Canada Secretariat, 2026).

So the federal market in a vendor's plan — one motion, many departments, compounding reference value — describes 24 companies out of 248. For the rest, federal revenue is a set of bilateral relationships that happen to share a payer, each won on its own.

Breadth is a channel property, not a product property

Look at what the widest suppliers carry, and the mechanism becomes obvious:

Supplier (merged)	Organizations	Value	Per relationship	What they carry
Microsoft	51	CA$201.18M	CA$3.945M	Client and application software licences
Northern Micro	25	CA$14.97M	CA$0.599M	Servers, desktop computer equipment
Nisha Technologies	15	CA$8.25M	CA$0.550M	Desktop computer equipment
Oracle	13	CA$41.73M	CA$3.210M	Software licences, operating-system maintenance
Carahsoft Technology	8	CA$1.12M	CA$0.140M	Software distribution

(Treasury Board of Canada Secretariat, 2026)

The second- and third-widest suppliers in the entire federal IT file are hardware resellers, not technology companies. Northern Micro reaches nearly twice as many federal organizations as Oracle does, on a third of Oracle's revenue. Carahsoft reaches eight organizations at CA$140,000 per relationship, which is what a distribution channel looks like from outside.

The lesson is precise. Breadth across federal organizations is bought by holding paper with many buyers, and holding paper is a commodity capability you can rent. A software company reaching one department directly plus eight through a distributor has the same coverage as one reaching nine directly, at a fraction of the fixed cost — and the disclosure file will record the distributor's name rather than yours, which is the trade.

Who the buyers actually are

Federal IT buying is more concentrated than federal IT spending. Across the 104 organizations, the top five hold 43.7% of contract value, the top ten 56.4%, the top twenty 72.0% (Treasury Board of Canada Secretariat, 2026). The Treasury Board of Canada Secretariat alone accounts for CA$330.32 million — 22.1% of the sampled total — largely on enterprise software licences bought once for use across government.

That last point catches product vendors. Some of the largest federal software purchases are made by a central organization on behalf of others, so the buying department and the using department differ, and a territory plan built on usage will call at the wrong door.

Two features that look like defects

The Financial Transactions and Reports Analysis Centre of Canada carries two vendor rows recorded as REDACTED, worth CA$26.35 million (Treasury Board of Canada Secretariat, 2026). A financial-intelligence agency withholding a supplier name is a legitimate feature of the disclosure regime rather than a data-quality failure — but it is CA$26.35 million no market-share analysis can attribute, and dropping it silently would overstate every named vendor's share.

And SAS Institute (Canada) Inc. sits at rank 1 with CA$203.23 million from a single organization, on two software licence and maintenance contracts awarded by the Treasury Board of Canada Secretariat on the same day in October 2024. It needs no merging; it outranks Microsoft. It is also present in exactly one organization, against Microsoft's 51 (Treasury Board of Canada Secretariat, 2026). Dollar rank and market presence answer different questions, and a slide showing only the first will send a sales team to the wrong doors.

None of this criticises any department or vendor. Proactive disclosure does what it was designed to do: publish awards so they can be looked up. The failure is treating a disclosure log as an analytical dataset, and that failure belongs to the analyst.

Where this fits in how I work

A competitive figure derived from a public file carries its merge rule plus its sampling frame in the same sentence as the number, before it reaches a client deliverable — because a reader who cannot see how supplier identity was decided cannot check what was counted Sagentix Phase 01 Market Intelligence, 2026. The deal-size distribution and the breadth distribution now go into every federal Phase 01 ahead of the competitor ranking, because they change the recommended motion and the ranking rarely does.

The full delivery system runs 6–8 weeks, draws on 1,412 curated artifacts, and is priced from CA$4,500 for Phase 1 to CA$45,000 for a full go-to-market build, depending on scope — with a Phase 1 money-back guarantee (subject to terms).

Three ways to act on this

Sort your vendor list by name, not by value, and read it. Value sorting hides duplicates by scattering them; alphabetical sorting stacks them. MICROSOFT CANADA INC, MICROSOFT CANADA INC. and Microsoft Canada Inc. land adjacent, and you will see them in seconds. Do this before any other cleaning step. It costs nothing, and it is the highest-yield pass on any procurement extract.

Replace your market-share slide with a deal-size histogram. Take the awards in your category, plot the distribution, mark where your current cost of sale breaks even. If the median award sits below that line — and at CA$338,316 across all federal IT it will, for most enterprise motions — your problem is motion design rather than competitive position, and no amount of share analysis will surface that.

Decide breadth or depth before you hire. Twenty-four suppliers in this file reach five or more organizations. If you intend to be one of them, budget for the paper — vehicles, registrations, a distributor relationship — because that is what the widest suppliers are really selling. If you intend to be one of the 162, build for a single deep account and stop calling it a federal strategy. A structured market-access pass earns its cost when that choice is about to be funded; it is a Phase 01 engagement, and I would take it after the first two, which cost an afternoon.

Canada publishes more procurement detail than most governments. That is a genuine advantage, wasted twice over — once when a disclosure log is read as a ledger, again when the only question asked of it is who is winning.

What is the merge rule behind the vendor names on your last competitive slide — and could you write it down right now?

References

• Treasury Board of Canada Secretariat. (2026). Proactive disclosure of contracts over $10,000 [Data set]. Government of Canada. Open Government Licence – Canada. Records re-derived 8 September 2026; each organization's 250 most recent contracts, 104 organizations carrying at least one information-technology award. Concentration figures are computed on the 573 named top-supplier rows; deal-size and category figures on the 578 most-recent individual award records.

Contains information licensed under the Open Government Licence – Canada. That licence covers the proactive contract disclosure dataset used throughout.