Statistics Canada - Manufacturers' Sales by Industry (incl. NAICS 33312)

It is Canada's official account of its own construction-machinery factories, kept monthly since January 1992. Table 16-10-0047-01 - CANSIM 304-0014 to analysts of an earlier vintage - is the published face of the Monthly Survey of Manufacturing: manufacturers' sales, inventories, orders and inventory-to-sales ratios tabulated across 351 NAICS industries, with Construction machinery manufacturing [33312] carried as member 235 in its own right rather than averaged away inside transportation equipment.

Four dimensions structure the cube, which is why it models so cleanly. Geography contributes exactly one member - Canada as a whole. Principal statistics contributes nine measures: shipments, new orders received during the month, unfilled orders at end of month, raw-materials inventories, goods or work in process, finished goods, total inventory, and the two inventory-to-sales ratios. Seasonal adjustment contributes two states, so adjusted and unadjusted readings sit side by side instead of in separate publications. NAICS contributes the 351 industries. Fully expanded, that is 2,840 live series and about 1,085,752 datapoints - roughly eight series per industry on average, with room for eighteen.

For anyone selling into or studying heavy equipment, the payoff is frequency with authority: a national statistical agency counting factory shipments, order books and inventories every month, reaching back further than most vendor series have existed. Get a sample of this dataset scoped to the measures and window you name.

Monthly, long-run and flat - one observation per series-month, keyed by reference date, measure, adjustment state and industry. Real rows for the NAICS 33312 cell, in delivered column order:

REF_DATE: 1992-01   GEO: Canada   adjustment: Unadjusted
naics: Construction machinery manufacturing [33312]
measure: Sales of goods manufactured (shipments)
value: 43,270   scalar: thousands of dollars   coordinate: 1.1.1.235

REF_DATE: 1992-01   measure: New orders, estimated values of orders received during month
value: 43,270   coordinate: 1.2.1.235

REF_DATE: 1992-01   measure: Unfilled orders, estimated values of orders at end of month
value: 27,413   coordinate: 1.3.1.235

REF_DATE: 2026-05   measure: Sales of goods manufactured (shipments)   value: 322,974
REF_DATE: 2026-06   measure: Sales of goods manufactured (shipments)   value: 368,793

The block tells the whole story the table is for. Unadjusted shipments ran $43.27 million in January 1992 and $368.79 million in June 2026 - better than eightfold across 414 consecutive months, no gaps. The newest month stepped up 14.2% over May's $322.97 million, exactly the kind of move you check against the seasonally adjusted twin before believing. And even in 1992, unfilled orders already stood at 63% of a month's shipments - order-book depth this series has tracked continuously ever since.

Seventeen documented fields define the row, with definitions verified against the cube's own metadata rather than inferred from documentation prose. Three do the analytical work: Principal statistics names the measure, Seasonal adjustment separates the twins, VALUE carries the number.

Four more keep you honest. SCALAR_FACTOR - values publish in thousands of dollars, the classic off-by-three-orders charting error waiting to happen. UOM, so units never get assumed. STATUS, which marks preliminary months that later revisions rewrite. TERMINATED, which flags discontinued series before you build on one.

Two identifiers ride along per row and earn their keep the moment anything automates: VECTOR is the stable series key (v800327 is unadjusted NAICS 33312 shipments) and COORDINATE encodes the dimension members dot-separated - 1.1.1.235 reads Canada, shipments, unadjusted, industry member 235. Everything above arrives with your sample by default; suppressed-value handling and any supplementary cuts ship on request.

Geography - Canada as one national total. There are no provincial members behind the NAICS 33312 series in this cube; regional cuts require a companion table, which we can scope into the same sample.

Temporal - January 1992 through June 2026, monthly, unbroken. The 33312 unadjusted shipments series carries 414 observations across that entire span, and the June 2026 reference month arrived in the 14 August 2026 release. A thirty-four-year monthly archive is long enough to train cycle models on several full booms and busts.

Granularity - one row per month x statistic x adjustment x industry. Across 351 NAICS industries that yields 2,840 series and about 1,085,752 datapoints, with the nine measures and two adjustment states giving up to eighteen series per industry.

The wider catalog this sits in is on our construction machinery heavy transportation equipment data hub.

API, files, or your warehouse. Daily, weekly, or hourly.

Pick the channel your team already works in: live queries pinned to specific series and windows, flat files sized for overnight loads, or a direct pipe into Snowflake, BigQuery or Redshift. Cadence is yours to set - and to change when your models change.

Every delivery ships with the full field dictionary, sample rows for validation, and series keys that hold steady across releases.

• Equipment-cycle timing - new orders, shipments and unfilled orders for NAICS 33312 are the accelerator trio: demand arriving, getting built, backing up. The natural home for this is our investors use cases page.
• Market sizing with provenance - officially counted Canadian factory shipments anchor TAM decks and attach-rate math instead of reconstructed estimates; see market researchers use cases.
• Channel-fill detection - finished-goods and total inventories against the two inventory-to-sales ratios show whether machines are accumulating faster than they ship.
• Forecast feature panels - a 414-month monthly panel with adjusted and unadjusted twins drops straight into feature stores; see data scientists use cases.
• Cross-border reads - Canadian shipments sit alongside US series for the same industry, splitting North American demand into clean domestic halves.
• Citable official figures - journalists and academics quote a national statistical agency's own counts, table number attached.

Investors and quant researchers rate this highest: monthly official measurements of the Canadian heavy-equipment cycle beat quarterly reconstructions every time demand turns. Market researchers and consultants come next for the same reason in reverse - when a market-size deliverable needs a denominator nobody can argue with, counted factory shipments are it. Data scientists get a three-decade monthly panel with stable series keys and both adjustment states. Journalists, academics and students cite figures whose provenance is a national statistical agency. Developers building data products load dimensional rows without reshaping, delivered daily, weekly, or hourly as they choose. Sales and competitive-intelligence teams use shipment momentum as the backdrop under territory plans and rivals' claims.

Four things worth having upfront.

First, geography here is Canada, full stop. The cube holds one geographic member, so any provincial angle on NAICS 33312 needs a companion table - name the provinces in your request and we scope that alongside rather than after.

Second, the money columns publish in thousands of dollars. Convert once, centrally, or not at all; the scalar factor travels on every row precisely so nobody guesses.

Third, recent months carry status flags. Preliminary estimates get revised, and revised again sometimes - keep the revision history rather than snapshotting a month the day it lands if your models are sensitive to it.

Fourth, two of the nine measures are ratios, not dollars, and suppressed cells arrive as nulls rather than zeros. Sum across measures blindly and you corrupt aggregates; treat ratios as ratios and nulls as unknowns and the panel stays clean. Say which measures you need and the sample arrives pre-cut to them.