Data Disclaimer | Taydle.com

Understanding Taydle Shipment Data

Last updated: May 15, 2025

This page explains how to read, interpret, and responsibly use the shipment‑level information available on Taydle.com and via TaydleGPT. Shipment manifest data is enormously valuable but also inherently messy. We invest heavily in cleansing, normalisation, and enrichment, yet the dataset will never be 100% perfect. The notes below highlight common data quirks, methodological choices, and the caveats you should keep in mind before drawing conclusions.

1. Nature of Shipment Manifest Data

Source‑of‑truth, not source‑of‑perfection – Bills of lading and customs entries are legal documents filed by thousands of companies, carriers, and brokers worldwide. Typos, abbreviations, and incomplete fields are inevitable.
Continuous updates – New records are ingested daily; historical records may be amended when a carrier or customs authority files a correction.
Sea‑freight focus – Public laws make ocean manifests transparent; air‑ and land‑cargo details are often unavailable. Absence of an air shipment in Taydle does not imply that the shipment never happened.

2. Product Search Caveats

Scenario	Why it Happens	How to Mitigate
False positives – a search for "green peas" returns a multi‑product consignment containing "yellow peas" and "green apples".	Manifest descriptions are free‑text; the search engine matches any record whose description contains the tokens you typed.	⬥ Review individual rows before bulk‑downloading. ⬥ Use exclusion terms (e.g. `green peas -apple`).
Synonyms & generic terms – "shirts" may be recorded as "garments" or "textile goods".	Shippers choose their own wording; there is no global vocabulary.	⬥ Query multiple synonyms. ⬥ Combine free‑text search with HS codes.
Partial disclosure – some manifests describe goods only at a broad level.	Exporters are not legally required to list every SKU.	⬥ Treat text search as directional, not exhaustive.

Take‑away: always experiment with several keywords and HS codes to build the fullest possible picture.

3. HS Code Classification Accuracy

HS Level	Typical Accuracy*	Remarks
Chapter (2‑digit)	High	Broad groupings (e.g. 08=Fruit, nuts) are usually correct.
Heading (4‑digit)	Moderate	Misclassifications increase; many multi‑product consignments use a single umbrella code.
Sub‑heading (6‑digit)	Lower	Detailed classification requires precise product specs often missing from manifests.

*Internal benchmarking across major trade lanes.

Why HS accuracy suffers

Single code, multi‑product – A container of shirts and shoes may be filed under one code.
Complex rule set – HS rules have exceptions even experts debate.
Free‑text limitations – Our machine‑learning model infers codes from descriptions, but gaps remain.

Practical use‑cases

Use HS codes for trend direction, filtering, and macro volumes.
Avoid relying on 6‑digit codes for precise market‑share unless validated with additional sources.

4. Company Identification & Deduplication

No global ID: Manifests lack a unique identifier for shippers / consignees.
Variant spellings: "Samsung Electronics Co.", "Samsung Electronics Co Ltd", "Sumsung Electroncis" may all refer to the same entity.
Our approach: Natural‑language clustering, fuzzy matching, and manual curation merge probable duplicates.
Residual error: Expect over‑counts or under‑counts when tallying unique exporters/importers.

Tip: when searching, start broad ("Samsung") then refine, and interpret metrics such as "# of exporters" as approximate.

5. Manifest Confidentiality & Redactions

Certain jurisdictions (notably the United States) allow companies to request confidential treatment. When granted:

Names and addresses of exporters/importers are replaced by placeholders in the public file.
Volumes attributed to the known players will dip when secrecy starts and spike when it lapses.
The "largest exporter" is therefore the largest known exporter.

Always cross‑check time‑series irregularities against confidentiality windows.

6. Multi‑Product Consignments

TEU/weight at container level: Weight, volume, and container counts apply to the whole box, not individual product lines.
Implication: Aggregating weight by keyword can overstate tonnage if a box contained multiple goods but only one matched your query.
Mitigation: Use shipment counts for market pacing; interpret weight & TEU as upper‑bound indicators.

7. Units & Measurement Issues

Unit mismatches: Rarely, a manifest states Unit=tons but value is in kg (×1000 error).
Our cleaning: Automated unit‑normalisation fixes the majority; some anomalies persist.
What you can do:
- Compare suspect values to typical density for that commodity.
- Exclude extreme outliers in your analysis.

8. Data Cleaning & Algorithm Evolution

Taydle employs machine‑learning models and rule‑based scripts that improve continuously. Re‑running the same report weeks later may yield slightly different counts as:

Late data arrives.
Erroneous records are corrected.
Our deduplication and classification models get smarter.

9. Responsible Use & Legal Disclaimer

Directional insights, not certified statements. The dataset is best suited for spotting trends, sourcing leads, and complementing – not replacing – official customs declarations or audited financials.
No warranties. Taydle disclaims all warranties, express or implied, regarding completeness, accuracy, or fitness for a particular purpose. Use at your own risk.
Verify critical findings. Before making high‑stakes decisions (e.g., contractual commitments, investment), validate with primary documents or direct supplier engagement.
Attribution & fair use. You may quote or visualise limited extracts with acknowledgement: "Source: Taydle.com shipment database." Bulk redistribution is prohibited without written permission.

10. Statistical Metrics Glossary

The summary view at the top of search results provides quick metrics to help you size markets and spot patterns. The table below defines each metric and flags the most important caveats.

Metric	Definition	Key Caveats
Number of Shipments	Total count of shipment records that meet your filters.	Cancelled or amended filings can create small over/under-counts.
Total TEU	Sum of declared Twenty-foot Equivalent Units across all selected shipments.	TEU for certain shipments could be incorrectly recorded. However, this is rare. TEU for LCL and Bulk shipments is set to zero. Value applies to the whole container even if only part of the cargo matches your query (see §6).
Total Weight	Sum of declared cargo weight (kilograms) for all selected shipments.	Subject to unit mismatches and rounding errors (see §7). Weight is for the entire consignment, not individual product lines.
Earliest Transaction Date	Date of the oldest shipment in the current result set.	Late filings or data corrections can push this date backwards. Confidential filings could impact this data.
Recent Transaction Date	Date of the most recent shipment in the current result set.	Processing lags mean the "current" date can be a few days behind real-time. Confidential filings could impact this data.
Number of Unique Suppliers	Count of distinct exporter entities.	Entity resolution is probabilistic; spelling variants may split or merge companies (see §4). Confidential suppliers are not included. Treat this as "known" suppliers rather than "total" suppliers.
Number of Unique Consignees	Count of distinct importer/buyer entities.	Same caveats as suppliers regarding name variants and confidentiality. Treat this as "known" consignees rather than "total" consignees.
Number of Unique Countries	Count of distinct origin countries in the selection.	Country field may be blank or reflect trans-shipment hubs rather than true origin.
Number of Loading Ports	Count of distinct seaports where cargo was loaded.	Trans-shipment ports can inflate counts.
Number of Discharge Ports	Count of distinct seaports where cargo was unloaded.	Feeder movements and in-land depots may be masked as the main discharge port.
Number of Unique HS Codes	Distinct Harmonised System codes appearing in the selection.	Subject to misclassification, especially at 4- and 6-digit levels (see §3).
Number of Unique Carriers	Count of distinct ocean-carrier SCAC codes present.	Feeder and slot-charter arrangements can mask the underlying carrier.
Number of Unique Vessels	Count of distinct vessels (by IMO or name) carrying the shipments.	Vessel renaming or code reuse may distort counts.
Avg Shipments per Supplier	Total shipments divided by number of unique suppliers.	Sensitive to both duplicate names and small sample sizes.
Avg Shipments per Consignee	Total shipments divided by number of unique consignees.	Same limitations as per-supplier metric.
Avg Weight per Shipment	Total weight divided by number of shipments.	Affected by missing or erroneous weight declarations (see §7).
Avg TEU per Shipment	Total TEU divided by number of shipments.	Same limitations as total TEU and Number of Shipments.
Number of Unique Dates	Count of distinct calendar dates on which shipments arrived.	Blank or corrected dates may under-count actual activity. Confidential filings could impact this data if used with shippers or consignees.
Number of Unique Months	Count of distinct calendar months represented in the data.	Same limitations as unique dates.
Number of Unique Years	Count of distinct calendar years represented in the data.	Long-run backfill of historical data can change this figure.

All metrics exclude records with missing or obviously invalid values. Figures are based on the public portion of the manifest dataset and should be treated as indicative, not definitive.

11. Feedback & Support

Found an anomaly? Need methodological clarification? Let us know at [email protected] – detailed data quality reports help us make the platform better for everyone.