Canada's Plans Show Federal IT Spending Falling. Actual Spending Has Beaten Plan Five Years Running.

The forecast that is sitting there for free

Every federal department publishes a Departmental Plan on or before 31 March each year, and inside it, for every programme, three years of planned spending and three years of planned full-time equivalents (Treasury Board of Canada Secretariat, 2026). Anyone can download the whole government's worth in one file.

For a vendor, that looks like the thing you would pay a research firm for: a named-buyer demand forecast, published, two to three years ahead of any tender.

So read it. Taking only the programme lines that are genuinely information technology — filed by the department itself under Internal Services, Common Government of Canada IT Operations, or an equivalent core responsibility, plus the mission ICT lines at Defence, the RCMP and the border agency — 36 organizations carry 85 IT programme lines with a forward plan, worth CA$5.08 billion in the plan year. The trajectory looks unambiguous:

Planned federal IT spending across the three plan years, 36 organizations

• Plan yearCA$5.08B
• Plan year + 1CA$5.03B
• Plan year + 2CA$4.54B

Thirty-two of the 36 organizations plan to spend less; of the four that plan to spend more, three also plan to add staff (Treasury Board of Canada Secretariat, 2026). On its face this is a shrinking market where the cut lands on purchases rather than payroll, and that is a genuinely useful thing for a vendor to know.

It is also mostly wrong, and the same file proves it.

The spine. A Departmental Plan is published before the fiscal year begins, so its out-year figures precede every in-year funding decision that follows. Check the plans against published actuals and the gap is large, one-directional, and stable: across 413 IT programme lines over five fiscal years, departments planned CA$17.87 billion and actually spent CA$22.41 billion — 25.4% more, with actuals beating plan in every single year and on 75.1% of individual lines (Treasury Board of Canada Secretariat, 2026). The apparent CA$539 million cut is an artifact of when the document is written. Most will read the plan as a demand forecast. That is the misread that will cost a vendor a quarter of its addressable federal IT market.

The control, on one programme

Start with a single line you can check by hand. The Canada Revenue Agency's Information Technology Services programme appears in three consecutive Departmental Plans, and two of them now have published actuals beside them:

Departmental Plan	Planned, plan year	Actual	Variance
2023–2024	CA$429.52M	CA$539.70M	+25.7%
2024–2025	CA$422.13M	CA$527.36M	+24.9%
2025–2026	CA$386.06M	not yet published	—

(Treasury Board of Canada Secretariat, 2026)

Two things are happening at once. Each plan forecasts a lower number than the one before — CA$429.52M, then CA$422.13M, then CA$386.06M — and each actual lands about a quarter above the plan it belongs to. A vendor reading the 2025–2026 plan sees a department scaling its IT programme down by 10% over two years. A vendor reading the actuals sees one that has spent over half a billion dollars in each of the last two measured years (Treasury Board of Canada Secretariat, 2026).

The plan is not lying. It answers one question — what has been approved by 31 March — and gets read as though it answered a second Sagentix GTM Methodology, 2026.

The pattern holds across the whole file

One programme is an anecdote. Here is every federal IT programme line where a plan-year figure and a published actual both exist:

Fiscal year	Lines	Planned	Actual	Variance
2020–2021	77	CA$2.01B	CA$2.83B	+40.9%
2021–2022	82	CA$3.25B	CA$4.22B	+29.6%
2022–2023	86	CA$3.98B	CA$4.74B	+18.9%
2023–2024	84	CA$4.24B	CA$5.35B	+26.2%
2024–2025	84	CA$4.39B	CA$5.27B	+20.2%
All five	413	CA$17.87B	CA$22.41B	+25.4%

(Treasury Board of Canada Secretariat, 2026)

Five years, five overshoots, a range of 18.9% to 40.9%. The median individual programme line spends 1.258 times its plan. Of 413 lines, 310 exceeded plan and 103 came in under (Treasury Board of Canada Secretariat, 2026) — so this is not a few outliers dragging an average, it is the ordinary case.

What would change my mind. A single year of actuals landing below plan across the file, or a mechanism that explains the gap as something other than in-year funding — a systematic re-mapping of programme lines between the plan and the results report, for instance — would break this. I checked the second by holding programme names constant across vintages, which is how the CRA row above was built. I cannot rule out re-mapping in every one of the 413 lines, and I would treat a single-line comparison more cautiously than the five-year aggregate.

Why the gap is one-directional

The mechanism I would offer is the timing, and I want to be clear that this part is my reading rather than something the dataset states. Departmental Plans are published on or before 31 March, before the fiscal year they describe has begun (Treasury Board of Canada Secretariat, 2026). Money that arrives after that date — supplementary estimates, budget measures, carry-forwards from a prior year, in-year reallocations — is absent from the document, because it does not exist yet.

That produces exactly the shape observed: a plan figure that is complete as at publication, an actual that includes everything added later, and a gap that runs in one direction because funding is far more often added than clawed back mid-year. It also explains why the out-years fall. Year two and year three of a plan can only contain money already approved for those years, and most of what will eventually be spent then has not been approved when the document goes to print Sagentix GTM Methodology, 2026.

The practical consequence is a rule. Read a plan-year figure as a floor. Read an out-year figure as a floor with more of the funding still missing. Never subtract one from the other and call the difference a cut.

What the plan is actually good for

Discarding the file would be the second mistake. The level is unreliable; the mix is not, because the same document that under-states the dollars also states how many people the department plans to have — and the staffing line does not move the way the spending line does.

Across the 36 organizations, 32 plan to reduce IT spending faster than they reduce IT staff. Fourteen of those plan to hold headcount within one percent of flat, or grow it, while the spending line falls. That difference is where a supplier's revenue lives, because a department that keeps its people and reduces its budget is reducing what it buys.

Organization	Planned spend change	Planned staff change	Dollar change
Shared Services Canada	−7.4%	−0.5%	−CA$94.59M
Public Services and Procurement Canada	−35.5%	0.0%	−CA$50.23M
National Defence	−3.9%	+2.1%	−CA$42.25M
Royal Canadian Mounted Police	−5.1%	0.0%	−CA$14.25M
Correctional Service Canada	−9.3%	−0.6%	−CA$9.75M
Innovation, Science and Economic Development Canada	−8.8%	+0.3%	−CA$3.29M

(Treasury Board of Canada Secretariat, 2026)

One organization is deliberately absent from that table. The Public Health Agency of Canada plans a CA$4.90 million reduction, but reports zero planned full-time equivalents against both of its IT programme lines in either year — so it has no staffing ratio to compute, and any percentage change from a base of zero would be an artifact rather than a measurement.

Public Services and Procurement Canada plans to keep every IT person it has and spend CA$50.23 million less. National Defence plans to add IT staff while its IT spending line falls CA$42.25 million. Whatever the absolute numbers turn out to be, the direction of substitution is stated by the department itself, and it is the same direction almost everywhere: hold the establishment, squeeze what gets bought Sagentix GTM Methodology, 2026.

For a vendor, that is a sharper signal than any spend total. It says the competitive threat in most federal accounts over the next two years is not another vendor. It is the internal team: in 14 of 36 organizations the headcount plan is flat or rising while the money plan falls, so your line item is the flexible one (Treasury Board of Canada Secretariat, 2026).

The two organizations that plan to grow

Four of the 36 plan a higher IT spend at the end of the horizon than at the start, and the amounts are small: the Canada Border Services Agency at +CA$8.46 million (+3.4%), Housing, Infrastructure and Communities Canada at +CA$1.08 million (+4.5%), the Canadian Space Agency at +CA$0.71 million (+4.7%), and the National Research Council at +CA$0.05 million (+0.09%) (Treasury Board of Canada Secretariat, 2026).

Given the +25.4% plan-to-actual gap, none of those is a reliable estimate of what will be spent. What they show is intent recorded in a public document at a moment when 32 peers recorded the opposite — which, on a target list, is worth more than the dollar figure attached to it.

Where this fits in how I work

Every forward-looking public-sector figure gets a plan-versus-actual check here before it reaches a client, and this one is why. The check took twenty minutes and turned a striking headline into an artifact of the appropriation calendar — which is the ordinary outcome of running a control, and the reason to run one on the number you like most.

Every Sagentix Phase 01 federal analysis now reports planned figures with their historical realisation rate attached, rather than as forecasts — because a public-sector demand number that has never been tested against outturn is a claim about a document, not about a market Sagentix Phase 01 Market Intelligence, 2026. 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).

This particular check needs no engagement. It is two columns in a file you can download this afternoon.

Three ways to act on this

Compute the realisation rate for your own target accounts. Pull the expenditure file, take the programmes you sell into, and for every year with both a plan-year figure and a published actual, divide one by the other. You will get a number per department. Multiply future plan figures by it before any of them reach a board slide or a territory model. For federal IT as a whole that multiplier is about 1.25; for your accounts it will differ, and knowing by how much is the entire exercise.

Stop treating out-year declines as demand signals. A plan's third year is structurally incomplete — it contains only money already approved that far ahead. If your pipeline model discounts an account because its plan shows a fall, you are modelling the appropriation calendar rather than the buyer. Model the level from actuals, and use the plan for direction only.

Read the staffing column beside every spending column. The plan's most reliable content is the ratio between the two. Where headcount holds and spending falls, expect procurement pressure, renewal scrutiny and insourcing conversations — and prepare the argument for why your line survives, before the renewal. That is worth a structured market-intelligence pass when several accounts move at once; it is a Phase 01 engagement, and I would reach for it after the first two, which cost an afternoon.

Canada publishes its buying intentions three years ahead, for free, in a machine-readable file. Almost nobody reads it, and the few who do read it as a forecast — which is the one thing it is not.

What realisation rate does your federal pipeline model assume — and has anyone ever checked it against outturn?

References

• Treasury Board of Canada Secretariat. (2026). Departmental Plans and Departmental Results Reports — Expenditures and full-time equivalents by program and organization [Data set]. GC InfoBase, Government of Canada. Open Government Licence – Canada. Re-derived 8 September 2026 from the CKAN datastore distribution, resource 64774bc1-c90a-4ae2-a3ac-d9b50673a895. Forward-trajectory figures use the 2025–2026 Departmental Plan vintage across 85 information-technology programme lines in 36 organizations; plan-versus-actual figures use all 413 information-technology programme lines from 2020–2021 to 2024–2025 that carry both a plan-year figure and a published actual.

Contains information licensed under the Open Government Licence – Canada. That licence covers the GC InfoBase expenditure dataset used throughout.