Blueship & Co.

A buyer abroad opens a directory, filters by category and city, and gets a wall of suppliers that all describe themselves the same way. The intermediary and the factory look identical in a listing. The buyer then spends a week sending emails, gets four replies, and still cannot tell which one owns machinery.

The usual business model makes that worse. When a platform is paid per enquiry it sells, its incentive is to maximise enquiries rather than matches, and supply-side ordering gets bought. The buyer never sees why one name sits above another.

A buyer sends four things: the product, the quantity, the destination and the date. We reply with a manufacturer from a shortlist we maintain and check.

Manufacturers list free — no listing fee, no ads, no credits. Nothing on the supply side can buy position, so a shortlist ends up ordered by fit: capability, capacity, paperwork. Not by who paid.

GSTIN, phone numbers and email addresses stay off the public page. That makes the individual manufacturer pages weaker for search and stronger for the thing that matters: a buyer cannot scrape the list, and a manufacturer does not get cold-called because they agreed to be listed.

The site states plainly that it is not a lead marketplace, and that enquiry volume is neither sold nor guaranteed. A sourcing site that promises enquiries is selling hope. One that promises a checked name is selling work.

The directory, the review queue and the enquiry pipeline are built in-house. Three surfaces: the public listing, an internal review queue, and the enquiry inbox.

The review queue is the interesting one, because it is a small state machine — submitted, checked, published — with a rejection path that has to carry a reason. A rejection with no reason is indistinguishable from a record nobody has looked at yet, and that ambiguity is how a directory rots.

Everything else is deliberately boring: a relational store, server-rendered pages, one queue. Throughput is not the constraint. I have to read every enquiry myself, so the software should cut what reaches me, not what leaves.

Verification is not a boolean and cannot be stored as one. A factory that can turn out five hundred units in a good month may not be able to make five thousand, and both statements are true of the same company. Capability is a range, capacity moves with the season, and the schema has to hold a range without rounding it into a lie.

Second: a shortlist is a judgement, and judgement does not scale by adding rows. The honest version of “we check manufacturers” is that a person checks them, and that person is the bottleneck. At this stage that is fine. At a hundred enquiries a week it would not be, and the answer will not be a bigger list.

Not a bigger directory. A review queue a second operator can run without me in the room: explicit criteria, an audit trail per manufacturer, and a shortlist that can explain itself to a buyer without a phone call.