📦 Factory Orders & Shipments API — Census M3 Manufacturing Data

Get the U.S. Census Bureau's M3 survey — Manufacturers' Shipments, Inventories, and Orders, the source of the monthly factory orders and durable goods reports — as clean records: value of shipments, new orders, unfilled orders, inventories by stage, inventory and unfilled-order ratios and monthly percent changes, for 87 industry categories, seasonally adjusted and not, monthly since 1992.

Economists, traders, manufacturers and analysts use M3 to read demand and inventories across US industry. Census publishes it as one file that refers to industries and data types by number; this actor joins those lookups so each record names the industry, the measure and its unit — no spreadsheet work, no API key.

The unit of output is one observation (one industry, data type, adjustment and month). You pay per observation delivered; a run that finds nothing costs nothing.

📊 Sample Output

period	category	data_type	value	seasonally_adjusted
2026-08	Total Manufacturing	New Orders	663491	true
2026-07	Total Manufacturing	New Orders	662837	true
2026-06	Total Manufacturing	New Orders	657790	true
2026-05	Total Manufacturing	New Orders	658836	true
2026-04	Total Manufacturing	New Orders	665895	true

Rows from live run o7trED7BQS47cbSt5 on Apify (input: categories MTM and NXA, data type NO, seasonally adjusted, 40 records). Notice the newest month first and the industry and measure named on every row (millions of dollars).

✅ What you get

Every field of each record:

• record_id — unique per observation: category, data type, adjustment, geography and month.
• period, period_label — month as 2026-08 and as published (Aug-2026).
• category_code, category — M3 category, e.g. MTM = Total Manufacturing.
• data_type_code, data_type — e.g. NO = New Orders, VS = Value of Shipments.
• error_type_code, error_type — sampling-error measures — empty for M3, which publishes none.
• unit_code, unit — e.g. MLN$ = Millions of Dollars, RATIO, PCT.
• seasonally_adjusted — true for seasonally adjusted rows.
• geography_code, geography — US = U.S. Total.
• value, value_as_published — the value as a number and exactly as published (Census asks that precision not be altered).
• data_updated_on — when Census last updated the file.
• program — the originating program name.
• source, source_dataset, licence, attribution, observed_at — provenance on every record: Census and the program, the file and its update time, the licence line with Census' attribution requirement quoted, the attribution, and the UTC read time.

⚙️ Sample inputs

1. Store example — total manufacturing new orders

{
  "categoryCodes": ["MTM"],
  "dataTypes": ["NO"],
  "adjustment": "seasonally_adjusted",
  "maxRecords": 10
}

The 10 newest months of seasonally adjusted new orders for total manufacturing; about $0.50 on the Free plan.

2. Narrowed — core capital goods orders and shipments since 2024

{
  "categoryCodes": ["NXA"],
  "dataTypes": ["NO", "VS"],
  "adjustment": "seasonally_adjusted",
  "periodFrom": "2024-01",
  "maxRecords": 200
}

Nondefense capital goods excluding aircraft (widely watched as a business-investment gauge): new orders and shipments from January 2024.

3. Bulk — latest month of every series

{
  "periodFrom": "2026",
  "latestOnly": true,
  "maxRecords": 3000
}

One record per series (industry × data type × adjustment): its newest month.

Input fields:

Field	Meaning
categoryCodes	M3 category codes, e.g. MTM, MDM, MNM, NXA, 36S
categoryContains	every word of an entry in the category name, e.g. aircraft
dataTypes	VS, NO, UO, TI, MI, WI, FI, IS, US and the MPC… percent changes
adjustment	both, seasonally_adjusted, not_seasonally_adjusted
periodFrom / periodTo	a month (2024-06) or a year
latestOnly	one record per series: its newest month
order	newest_first (reads the small file to the end) or oldest_first (streams and stops at the cap)
maxRecords	most observations delivered and billed (default 1,000)

🧾 JSON sample record

{
  "record_id": "M3:MTM:NO:SA:US:2026-08",
  "period": "2026-08",
  "period_label": "Aug-2026",
  "category_code": "MTM",
  "category": "Total Manufacturing",
  "data_type_code": "NO",
  "data_type": "New Orders",
  "error_type_code": null,
  "error_type": null,
  "unit_code": "MLN$",
  "unit": "Millions of Dollars",
  "seasonally_adjusted": true,
  "geography_code": "US",
  "geography": "U.S. Total",
  "value": 663491,
  "value_as_published": "663491",
  "data_updated_on": "Friday, 02-Oct-26 09:02:34 EDT",
  "program": "Manufacturers' Shipments, Inventories, and Orders (M3)",
  "source": "U.S. Census Bureau — Manufacturers' Shipments, Inventories, and Orders (M3) Survey (economic indicator time-series file M3-mf.csv)",
  "source_dataset": "M3-mf.zip (M3-mf.csv) from census.gov/econ_getzippedfile/?programCode=M3, data updated Friday, 02-Oct-26 09:02:34 EDT",
  "licence": "U.S. government work, not subject to copyright in the United States (17 U.S.C. § 105). Census Bureau file README: \"All products developed using these data files must attribute the datasource as the U.S. Census Bureau and must contain the originating program name.\"",
  "attribution": "Source: U.S. Census Bureau, Manufacturers' Shipments, Inventories, and Orders (M3) Survey",
  "observed_at": "2026-10-04T05:12:33Z"
}

🔧 How it works

1. Robots check. www.census.gov/robots.txt is read on every run; its rules for all crawlers disallow only /search-results.html, so the file path is allowed. A real disallow would stop the run with nothing charged.

2. Streamed file. The actor opens Census' download link census.gov/econ_getzippedfile/?programCode=M3 and decompresses the ZIP as it arrives. The lookup sections at the top of the file (categories, data types, geography, time periods) are read first; the data rows follow and are joined to them, so records carry names, not index numbers.

3. Filter, deliver, charge. Rows are checked against your categories, data types, adjustment and month window. With oldest_first the actor delivers as it reads and stops at maxRecords; with newest_first (the default) it reads the whole file — it is 2.6 MB compressed and takes seconds — then delivers each series newest month first. Every batch is charged after it is delivered.

4. Receipt. RUN_RECEIPT records the file, Census' "data updated on" time, rows read and matched, and delivered and charged counts.

No login, API key, proxy or bot workaround — one public Census file, and no personal data in it.

💰 Pricing example

Per observation delivered, no start fee, no monthly charge:

Apify plan	Price per observation
Free	$0.05
Bronze	$0.045
Silver	$0.04
Gold, Platinum, Diamond	$0.0335

• 10 observations (the Store example): $0.50 on the Free plan, $0.335 on Gold.
• 1,000 observations (the default cap): $50 on Free, $45 on Bronze, $40 on Silver, $33.50 on Gold.

Coverage and size. Live count from the file read on 3–4 October 2026 (M3-mf.zip, 2.57 MB compressed, M3-mf.csv 11.6 MB; data updated Friday, 2 October 2026 09:02:34 EDT): 603,178 observations — 87 industry categories, 16 data types, U.S. total, January 1992 to August 2026, half seasonally adjusted (301,589) and half not (301,589). The planning index counted 603,178 rows — they agree. A filter is required (categories, a name word, data types or a start month); a whole-file read takes seconds, and a full pull is 603,178 records, billed per observation.

⚖️ Legal & ToS

Data: U.S. Census Bureau, Manufacturers' Shipments, Inventories, and Orders (M3) Survey, economic-indicator time-series file. Works of the U.S. government are not subject to copyright in the United States (17 U.S.C. § 105). The README inside Census' file says: "All products developed using these data files must attribute the datasource as the U.S. Census Bureau and must contain the originating program name." Every record carries that attribution and the program name.

The actor reads only this public file at Census' published download link with a declared User-Agent; www.census.gov/robots.txt allows the path and is checked on every run. No login, key, proxy or form. Industry aggregates only — no personal data (the README's contact address is never copied).

This actor is not affiliated with or endorsed by the Census Bureau. Current-month M3 figures are preliminary and are revised in later months, as Census notes in the file.

❓ FAQ

Do I need a Census API key? No. The actor reads Census' public download file, which needs no key or registration.

Why must I give a filter? The file holds 603,178 observations; a filter keeps runs fast and bills only what you asked for. An unfiltered run is refused with a clear message.

Is this the durable goods report? It is the M3 survey behind both the advance durable goods report and the full factory orders report: durable goods are category MDM, core capital goods NXA.

How fresh is the data? As fresh as Census' file: every record carries Census' "data updated on" time. The latest month is preliminary and revised later.

Why are some values percentages or ratios? Data types IS and US are ratios and the MPC… types are monthly percent changes; the unit field says which.

🆘 Troubleshooting

• 0 records — category codes are short codes such as MTM or 36S; check the data type code (e.g. NO).
• Both adjusted and unadjusted rows — set adjustment to seasonally_adjusted or not_seasonally_adjusted.
• Oldest months came first — that is oldest_first; use the default newest_first.
• SOURCE UNAVAILABLE message — the server did not answer after three tries; nothing was charged — run again later.
• CAPPED message — your maximum total charge was reached; raise it in the run options.

🔗 Related Actors

• Sweden Industry Turnover Monthly - SCB — monthly industry turnover in Sweden.
• Eurostat Business & Consumer Confidence — Monthly — European business and consumer confidence.
• UK GDP Four-Decimal Time Series - ONS — UK GDP time series.
• Sweden GDP by Industry & Quarter - SCB — GDP by industry for Sweden.

🏢 About NexGen Signal — official-source data products from government and regulator publications. Browse the catalog: apify.com/nexgensignal

⭐ Found this useful?

If this actor saved you time, a rating on its Apify Store page helps other buyers find it.

Part of NexGen Signal — official-source data products.