How to Shortlist Jewelry Manufacturers Before You Contact Anyone

The usual state after a fortnight of searching is a spreadsheet with thirty supplier names, a column of website links, and no basis for putting any of them above the others. Contacting all thirty produces a second problem: thirty conversations, most of them wasted, and no capacity left for the few that matter.

Below we set out how to cut a long list to a working shortlist using only what is visible before contact, so that the enquiries you do send go to candidates worth the reply.

Sorting is a different job from searching

Discovery and evaluation use different evidence and different attention, and mixing them is why long lists never shorten. While you are still adding names, every candidate looks plausible; while you are ranking, every candidate looks flawed.

Close the search first. Set a number of names you consider enough, reach it, and stop. Then rank what you have with the search tab shut. The discipline matters because ranking against an open-ended list is impossible — there is always one more supplier to check, and the shortlist never forms.

The disqualifiers to apply first

Start by removing, not by scoring. Removal is faster and less arguable, and a long list usually loses a third of its entries to four checks.

• Wrong metal. A workshop whose visible work is outside 925 silver and 14 or 18 ayar gold is not a fit for a precious metal line, whatever it says it can do.
• Wrong category. No evidence of your product type — cast pendants, chain assembly, personalised work, set rings — anywhere in their output.
• Wrong order shape. Everything visible is single-style volume when your range is varied, or the reverse.
• No traceable business. No address, no company name, no register entry, no consistent identity across the places they appear.

Score what remains on evidence, not on presentation

A polished website is a marketing decision, not a manufacturing one, and scoring it rewards the wrong businesses. Score instead on the kind of evidence available before you speak to anyone.

Three tiers are worth distinguishing. Product photography alone is the weakest — it can belong to anyone. Images or footage of work in progress, showing hands, tools and a room, is materially stronger. Third-party traces — an association register entry, a fair exhibitor listing, a brand publicly naming them as a production partner — are strongest, because someone other than the supplier put them there. Give the third tier the most weight. A candidate with two independent traces beats one with a better gallery.

Read the business model from what is visible

You are trying to establish, before contact, whether a candidate makes goods or places orders. It cannot be settled at this stage, but it can be estimated, and the estimate should affect the ranking.

The signals are ordinary. A catalogue spanning wildly unrelated categories suggests procurement. A narrow, coherent body of work in one metal family suggests a bench. Photographs that repeat across several supposedly different companies suggest a shared source. A workshop describing which operations it sends out is more likely making than one describing only what it can supply. Determining this properly is a later stage with its own methods, and our Turkish note on telling an intermediary from a manufacturer covers them.

Check the constraints that no amount of quality fixes

Some candidates fail on facts that have nothing to do with how well they make things, and those facts are cheaper to check now than after a sample.

Does the workshop export at all, or only sell domestically? Is there anyone who works in a language you can write a specification in? Can they ship to your destination, and have they done so? Does their visible order shape include anything as small as your first order is likely to be? A candidate failing any of these is not a bad workshop; it is a workshop for someone else, and it should leave the list without a conversation.

Rank into three groups, not into a numbered order

A ranked list of fifteen invites you to work down it, which spreads effort thinly across candidates you have already judged weaker. Three groups work better.

Put a small number — enough to sample from, few enough to manage properly — in the first group and give them your full attention. Put the plausible remainder in a second group, held without contact, to be promoted if a first-group candidate fails. Everything else goes in a third group that you do not revisit. Write one line beside each second and third group entry saying why it is there, because the reason fades within weeks and you will otherwise re-research the same names.

What to leave for after contact

Resist finishing the evaluation on paper. Pricing, minimum order thinking, sample quality, communication speed and the willingness to answer an awkward question are all post-contact evidence, and they overturn desk rankings regularly.

The deeper verification work — documents, live video of the space, tests performed on a physical sample — belongs after the shortlist exists, not before. Doing it on fifteen candidates is how a sourcing project stalls; our Turkish guide to verifying a manufacturer remotely is the stage that follows this one.

The rule for closing the list

A shortlist is finished when you can state, in one sentence each, why every remaining candidate could plausibly win. Not why they are on the list — why they could beat the others.

If you cannot write that sentence for a candidate, they are filler, and filler is what turns a shortlist back into a long list. Cut them and keep the number small enough that you would genuinely buy a sample from every name remaining. That is the real test: a shortlist you would not sample in full is not a shortlist. When you are ready to write to the first group, our manufacturing page and contact page show what we need in a first message.