Prospecting
How to find export buyers using shipment data
Directories give you companies that might buy. Shipment records give you companies that already did, last month, in a quantity you can read. This is the method we use, with the arithmetic shown.
Most export prospecting starts in the wrong place. Someone buys a directory, exports 4,000 rows of "importers of ceramic tiles in Europe", and the team spends six weeks emailing companies that turn out to be freight forwarders, retailers who buy domestically, or businesses that stopped importing in 2019. Reply rate lands near zero. Everyone concludes cold outreach is dead.
Cold outreach is fine. The list was the problem. A directory records who says they are in a business. A shipment record proves who moved goods, with a date, a quantity, a port and a counterparty attached. Those are two different populations of company, and only one of them can be qualified before you spend money on contact.
Your buyers sit inside a handful of those 11,565 lines.
What a shipment record actually contains
Two very different documents both get sold as "trade data", and they carry different fields. Knowing which one is on your screen matters more than most people realise:
| Field | Customs declaration | Bill of lading / manifest |
|---|---|---|
| Exporter / shipper name | Yes | Yes |
| Importer / consignee name | Yes (export side) | Yes, unless "TO ORDER" |
| HS code | Yes, 8 digits | Sometimes, often 4 to 6 |
| Declared value | Yes, in INR | Rarely |
| Quantity and unit | Yes | Weight and container count |
| Ports, vessel, container no. | Ports only | All of it |
| Address / contact trail | Limited | Often present |
Neither is complete on its own. Customs data gives you value, which is what any pricing work needs. Manifests give you logistics and identity, which is what outreach needs. A serious prospecting pass reads both. We wrote up the differences and the gaps in what is actually in India's trade data.
The method
-
Pin your product to an 8-digit code, not a 4-digit one
Everything downstream keys off this. Chapter 69 is "ceramic products", worth roughly $2bn of Indian exports a year. Heading 6907 is tiles.
69072100is glazed tiles with a water absorption of 0.5% or less, which is a specific, quotable product. Search at chapter level and your buyer list comes back thirty times too big and 90% irrelevant.Don't take the code from an internal spreadsheet. Verify it, then check it against what your competitors actually file under. The two disagree more often than people expect, and we go through why in the ITC-HS guide.
-
List every party that imported that code, and when
Filter shipments on the code, then on destination country if you already have a market in mind. What comes back is a list of consignees with a shipment count against each. That is your raw universe, and it is usually a few hundred names per code per country rather than thousands.
Keep the date window tight. Twelve months is the right default: long enough to survive seasonality, short enough that everyone on the list is still trading.
-
Rank on frequency, not size
Most people get this backwards. The biggest single shipment on the list is usually a one-off: a project order, a relocation, a trader covering a gap. The company that filed fourteen shipments in twelve months has a running programme, a budget line, and a procurement person whose job is to keep it supplied.
A buyer who imports monthly will take your call. A buyer who imported once, eight months ago, will not.
-
Read the incumbent supplier
Every one of those shipments names who shipped it. That gives you three things at once: who you are competing against, which country they ship from, and roughly what the buyer is paying, from declared value divided by quantity.
The pitch then changes depending on the answer. If the incumbent is Chinese and the buyer is European, tariff and lead-time arguments are live. If the incumbent is another Indian exporter twenty minutes from your own factory, you are competing on price and reliability alone. Better to know that before you pick up the phone than halfway through the call.
-
Score the shortlist before you contact anybody
Contact costs money and attention. Score first, using evidence you already have:
A scoring rubric that only uses fields present in the shipment record. Score each buyer 0 to 10, then work the top decile first. Signal Weight What "good" looks like Shipment frequency ×3 6+ shipments in 12 months Recency ×3 Something in the last 90 days Trend ×2 Second half of the year bigger than the first Supplier concentration ×2 3+ different suppliers, so they switch Price band ×1 Unit value at or above your quotable price Country fit ×1 You can actually serve the lane Supplier concentration is the one worth dwelling on. A buyer who has used the same supplier for three years is loyal and expensive to move. A buyer running three or four suppliers has already decided multi-sourcing is their policy. They have a slot open, and filling it is a commercial conversation rather than a rescue mission.
-
Get to a named person, then write like you have read the file
A company name is not a prospect. The shipment record's address, the company registry and the buyer's own site get you to a procurement or import manager. After that the email almost writes itself, because you know their code, their volume, their lane and their incumbent. That first specific line is where the reply rate comes from, and we go through the structure in the cold email post.
What the funnel actually looks like
Below is the shape you should expect from one code in one country, over one afternoon. The figures are illustrative and yours will differ, but the ratios hold across most industrial categories we have looked at.
Forty-one qualified accounts is a quarter of work for one salesperson. Three hundred and twelve unqualified ones is a quarter of work for nobody, because a list that size never actually gets worked.
Three mistakes that quietly ruin the list
Treating a forwarder as a buyer
Freight forwarders, customs brokers and consolidators turn up as consignees constantly. They are not buying your product, they are moving somebody else's. The tells are obvious once you look for them. The name contains "logistics", "forwarding", "shipping" or "cargo". The same party appears across twenty unrelated HS codes. The shipment mix makes no commercial sense for one business. Strip them out before you score, or they will sit at the top of your list by volume and cost you a week.
Reading a "TO ORDER" consignee as a real company
On a negotiable bill of lading the consignee field often reads TO ORDER or
TO THE ORDER OF [bank], because the cargo is released against endorsement of the
document and no name gets printed. That is a legal instrument, not a customer called "To Order".
Any dataset that ranks it among the top importers of a product has not cleaned its inputs, which
tells you something useful about the rest of it. More on this in
how to read a bill of lading.
Comparing declared values across currencies and units
Indian customs values are filed in INR. Your buyer's procurement team thinks in USD or EUR. Quantity units vary line to line, sometimes square metres, sometimes pieces, sometimes kilograms for the same product. A unit value only means something against another unit value in the same currency and the same unit. Get it wrong and you will confidently misquote by 40% in one direction or the other. Building the comparison properly is covered in export pricing and the market band.
No Indian trade dataset, ours included, is a complete census of every shipment through every port. Coverage varies by port, by year and by direction, and the import side is thinner than the export side. So a nil result means "not in this data". It never means "this company does not import". Treat an absence as a question to go and ask, not an answer you have found.
Where this method stops working
I would rather be clear about the boundary than oversell the technique. Shipment data finds buyers who already import your product category by sea or air. It will not find:
- buyers who source domestically today and might switch to imports, since there is no record of a shipment they never made;
- most intra-EU or overland trade, which never generates a bill of lading;
- the real decision-maker behind a group that imports through one procurement subsidiary in a third country;
- services, licensing, and anything moving under courier consolidation.
For the first case, look-alike modelling is the right tool and a genuinely different one. You take a buyer you have already qualified and find businesses in the same country and the same line of trade, whether or not they have filings. For the rest, trade fairs and channel partners still earn their keep. The data has a narrower job there, which is telling you which fair is worth attending.
A one-week version
- Monday: confirm your 8-digit ITC-HS code against the DGFT schedule, and against what competitors file.
- Tuesday: pull importers of that code for two target markets, twelve-month window. Strip forwarders and TO ORDER rows.
- Wednesday: score the list. Keep everything at 7+ and park the rest.
- Thursday: for the top 20, read the incumbent supplier and the unit value. Write one line of context per account.
- Friday: find named contacts and send 20 individually written emails, not 400 templated ones.
- Monday after: follow up on non-replies once, then start again with the next 20.
Twenty properly researched emails a week comes to about a thousand a year, every one of them aimed at a company that demonstrably buys your product. Start next Monday with one code and one country. The list will be smaller than you expect, which is rather the point.
