China industrial supplier directory: what the listings layer holds

A China industrial supplier directory answers the buy-side question every other record in this industry dodges: who in China makes or moves this component, at what indicative ask, in what order quantities, and can the company behind the offer be verified. Of the twenty records pooled under industrial machinery & supplies & components - seven primary, thirteen cross-listed neighbours - exactly one records suppliers themselves: Made-in-China.com, cataloged by Datadory at 7 out of 10 on our rubric against an all-catalog average of 7.81 across the 1,744 datasets we track. The other nineteen measure demand cycles, productivity or failure physics. Not one of them names a seller.

Seven verified fields carry every row, each defined against an observed listing during the August 2026 verification pass rather than inferred from a column header. product_title keeps the seller-written headline - bearing type, brand references and application claims packed into one string. price_range_usd holds the indicative dollar band shown on the listing card, a single figure when no range is quoted. moq preserves the order floor with its unit vocabulary ('1 Piece', '50 Pieces', '100 Sets') instead of flattening counts into one type. supplier_subdomain, supplier_name, supplier_location and audited_supplier_badge describe who stands behind the offer: the storefront key that survives individual offers churning beneath it, the registered legal name ready for registry and vendor-master matching, the province-level location, and the third-party verification flag.

Rows arrive flat, exactly as a delivery resolves them:

```text # one row = one product listing (bearings vertical, August 2026 verification pass)

Read the anatomy before the digits. The storefront key does the joining: every listing Changzhou Nanyi Bearing publishes shares yongjie-bearing even as individual offers appear and vanish, so a directory keyed on the storefront stays stable while its contents churn. The audited flag is the screen that reorders every downstream answer - all five rows captured in the verification pass carried it, which is typical of export-oriented storefronts and exactly why unfiltered averages mislead.

Geography - suppliers based across China's component-manufacturing provinces selling to buyers worldwide, with Liaoning, Shandong and Jiangsu anchoring the verified sample. Province is the geography Chinese component manufacturing actually clusters by: a bearing question is very often a Shandong question.

Temporal - a live marketplace floor. Listings reprice, relist and expire on the seller's schedule and no historical archive sits behind the board, so price history accrues from repeated captures on whatever rhythm you set. Weekly deliveries build a usable price-and-assortment curve inside a quarter; hourly catches repricing windows weekly snapshots miss. Delivered daily, weekly, or hourly - your call.

Granularity - one record per product listing, grouped by supplier storefront and category, roughly thirty listings per search-results view, and no precomputed rollups: every downstream cut aggregates from the atomic row, which is what keeps results reproducible. Scale runs to millions of live listings across thousands of storefronts under a category tree spanning industrial equipment and components, manufacturing and processing machinery, tools and hardware, auto parts, and electrical and electronics.

The honest caveat rides with every number: these are asks displayed on cards, not transactions. Negotiation happens below the printed band, so treat any derived price index as directional and let the spread between bounds say more than any midpoint.

A directory is not a pile of listings; it is a filtered, keyed and ranked subset. Five steps take the rows there:

1. Fix the category envelope first. Name the branches - bearings, transmission components, material-handling equipment - before any other filter, because titles are written by sellers as matching bait and bury the taxonomy in sales copy.
2. Collapse on the storefront key. Grouping by supplier_subdomain turns thousands of scattered offers into one row per company - the unit every outreach list and vendor master actually wants.
3. Stratify on the audited flag. Third-party verification separates verified companies from the rest; comparing price bands without holding that screen constant mixes two different markets.
4. Rank on quoted-band midpoint against the MOQ floor. Price range beside order minimum is the pair that converts browsing into per-unit economics, and it ranks candidates faster than citing folklore price points.
5. Compare like vintages only. Because the board keeps no back-catalogue, successive dated deliveries are the trend line: a revised or vanished card lands as a fresh vintage of the same row rather than a stale cached copy.

Category-tree placement, photography references and storefront-level rollups travel as additional fields on request - pinned against live records when your sample is cut rather than promised blind, so the schema you evaluate is the schema you ship against.

Directory fields establish who exists; association statistics establish whether the market they sell into is expanding. Five records bracket the listings layer from above.

JMTBA Machine Tool Statistics (Japan) publishes monthly machine-tool orders split domestic versus foreign from an archive reaching back to January 2009. The July 2026 flash printed 193,102 million yen - a month-on-month index of 94.9 against a year-on-year index of 150.4 - inside a 2026 cumulative of 1,248,243 million yen running at a year-to-date index of 137.8. A supplier feeding Japanese machine-tool exports sits inside that curve.

VDMA Mechanical Engineering Industry Statistics contributes the European leg: German order intake with domestic and foreign splits, alongside capacity utilisation of 77.1 percent in mechanical engineering as of January 2026 and a slightly positive H1 2026 order book carried by foreign business.

The American demand clock runs through Census M3 Manufacturers' Shipments, Inventories and Orders - roughly 600,000 rows of monthly shipments, new orders, unfilled orders and inventories since January 1992 - sharpened for metalworking by AMT USMTO, whose June 2026 print of $672.7 million (+56.8 percent year on year) lands inside a record $3.44 billion first half.

For structural benchmarks, the NBER-CES Manufacturing Industry Database contributes the 1958-2018 annual panel across 364 six-digit NAICS industries - the long-run reference for judging whether a quoted machinery ask is plausible against US productivity and capital intensity. Cross-border corridors come from Trade Map (International Trade Centre - ITC): roughly 220 reporting territories against about 5,300 HS6 products with values, quantities, unit values and tariff indicators, annual series generally from 2001. Filter reporter and partner and the shortlist acquires a trade-flow spine.

Yes - at the level of physics rather than vendors. The NASA Prognostics Data Set Repository serves 21 run-to-failure archives - turbofan degradation, bearing vibration and accelerated degradation, milling tool wear, battery aging, IGBT and capacitor aging - sized from about 12 MB to 15.8 GB, each carrying true remaining-useful-life labels beside per-cycle sensor readings. When a bearing supplier claims a wear life, these archives define the failure trajectories your acceptance testing should model. The Kaggle mirror of C-MAPSS is the community-standard entry point: four subsets, FD001 through FD004, totalling 25,888,576 bytes uncompressed, with FD001 alone contributing 20,631 labelled sensor-cycle rows.

No directory field certifies a part, and none pretends to. What the pairing changes is the question asked of a supplier: from trusting a verification badge to testing against labelled degradation data. The audited flag keeps the company-level screen honest; the engineering archives handle the component-level one.

• Sales and growth teams land at relevance 3: the storefront universe works as an account base, with audited status, MOQ band and province filtered into ranked outreach lists instead of trade-fair business-card stacks. The workflow expands at sales & growth teams use cases.

• E-commerce operators, also relevance 3, source components and compare quotes, order minimums and audit status side by side before committing purchase orders - the buying workflow itself, mapped at e-commerce operators use cases.

• Market researchers and consultants (relevance 2) map which provinces and companies cluster around a category - bottom-up scaffolding for market-entry and sizing work, continued at market researchers use cases.

• Competitive intelligence and product teams (relevance 2) track rival products' China ask prices, order minimums and badges across successive captures; shifts surface here before annual reports admit them (competitive intel use cases).

• Data scientists and ML engineers sample listing-price and MOQ distributions for China supply-side studies on a consistently shaped corpus (data scientists use cases); developers and builders wire supplier lookups into procurement and CRM tooling on a stable per-listing schema (developers & builders use cases); investors and quants gauge how broad and how cheap a China-exposed industrial's domestic supply base runs before underwriting its margins (investors & quants use cases); and journalists and academics cite named suppliers and observable asks when China's export manufacturing makes news (journalists & academics use cases).

Because the hard part was never finding a listing - it is turning millions of prose cards into rows that survive a join. Titles bury category in sales copy; price bands arrive as ranges with currencies attached; order minimums ride as bare counts whose unit vocabulary matters ('100 Sets' is not '100 Pieces'); province spellings drift between transliterations; and the audited flag that reorders every comparison sits one attribute deep on the card. Each of these is survivable once. None is worth re-solving every sourcing cycle.

Start with the industrial machinery supplies components data guide, the pillar scoring all twenty pooled records by quality, coverage and workflow. The nearest sibling clusters: machine tool orders japan monthly for the demand cycle behind the listings, glass bottle manufacturers China MOQ for the same listings logic applied container by container, and China origin component risk score for screening a bill of materials once suppliers are shortlisted. When the choice is telemetry versus order books, the Kaggle C-MAPSS versus JMTBA comparison settles it row by row.

Product-level detail - sample rows, field dictionary, coverage chips - lives on the Made-in-China.com page, and the industrial-machinery data hub indexes the full pool beside the best industrial machinery datasets ranking. Open the record, read the dictionary, then request a sample scoped to your categories and provinces.