Trade data vendors are not in the habit of publishing their own limitations. This is that page, because knowing where a dataset ends is the difference between using it well and quietly drawing wrong conclusions from it for a year.
What the records genuinely contain
Every ocean shipment entering the United States generates a bill of lading filed with US Customs and Border Protection. Those filings are a public record, and they carry:
- The consignee — the US company receiving the goods, with an address
- The shipper — the overseas company sending them, with a country
- A goods description — free text, written by whoever filed
- The date, the port, the container count and the weight
- An HS code — the tariff classification claimed for the goods
Aggregate that across years and you can answer some genuinely valuable questions: which US companies buy this product, who supplies them, how often, whether the relationship is growing or fading, and which factories a competitor uses.
That is a lot. Here is what it is not.
The gaps that matter
Air freight is mostly missing
The filing regime that makes this data public covers ocean cargo. Air shipments are not captured the same way. For a product that mostly flies — anything small, high-value or urgent, which includes a great deal of electronics, pharmaceuticals and samples — the records will understate the market badly, and a company that imports exclusively by air can be invisible.
Any company can ask to be hidden
US shippers can file a confidentiality request to have their identifying details withheld from the public record. Plenty do, and larger companies do it more often than small ones. When a company has done this, its shipments still exist and still count in totals, but its name does not appear.
The practical consequence: absence from your results is not evidence of absence from the market. If a company you know imports your product is missing, the likeliest explanations are a confidentiality filing, an air route, or a goods description you have not thought to search for — in roughly that order.
It is the United States
Everything above is about US imports. Other countries run their own regimes with their own coverage and their own rules; some publish nothing at all. A dataset built on US filings tells you about the US market and makes no claim about Europe or Japan.
The consignee is not always the buyer
The company named on the filing is whoever the container was shipped to. That is frequently a freight forwarder, a customs broker, a bonded warehouse or a trading intermediary rather than the company that will actually use or resell the goods. Filtering these out is real work, not a detail.
The goods description is free text
It was written under time pressure to satisfy a customs officer, not to describe a product to a buyer. It is abbreviated, inconsistent, frequently in capitals, sometimes wrong. Two identical shipments from the same factory to two different importers can be described in ways that share no words.
HS codes are claimed, not audited
The code on a filing is the one the importer’s broker chose. Brokers choose codes that are defensible and cheap, in that order. Similar products get filed under different chapters routinely, and the code is a strong hint rather than ground truth.
Questions the data cannot answer
Worth stating plainly, because these get asked of it anyway:
“What did they pay?” — Value fields are unreliable where they exist at all. Shipment count and weight are proxies for spend. They are not spend.
“Are they unhappy with their current supplier?” — The records show a relationship exists and roughly how active it is. A relationship that is shrinking is a hint. It is not a signal of intent to switch, and treating it as one produces confident emails to companies that are perfectly happy.
“Is this the whole market?” — No. It is the part of the market that moves by sea, into the US, without a confidentiality filing. That is a large and useful part. It is not all of it.
“What will they buy next?” — Nothing here is forward-looking. Every record is a thing that already happened.
How to use it well anyway
None of this makes the data weak. Used properly it is the strongest prospecting signal available for physical goods, because it is evidence of a completed purchase rather than a stated interest or an inferred firmographic.
The way to use it well is to treat a result as a company worth investigating rather than a qualified lead:
- Use activity in the last twelve months, not lifetime totals
- Check what share of a company’s imports are actually your product
- Verify the company exists and which website is theirs before writing
- Assume the list is incomplete and that the gaps are not random
- Never present someone’s shipment history back to them as though you have been watching — reference the category, not the file
Do that and you have something no lead database can match: a list of companies that have demonstrably bought what you make, from someone else.
Portaxo is built on these records and says so where they run out — including flagging companies that have filed for confidentiality rather than filling the gap with an estimate. See what it finds for your product, free.