The short answer
Before any bulk order, ask for four things in one pack: the physical samples themselves, a written spec sheet for each, the brewing or cupping parameters the packer used, and the documents that prove what the sheet claims - origin and batch records at minimum, lab panels where your market needs them. Then cup the samples against a reference you control, on a fixed protocol, and record the result. A sample that arrives without paperwork is not a sample; it is a flavor opinion in a bag. The whole point of the round is to produce a written, mutually held definition of the tea you are buying, so that the bulk lot can be checked against something more durable than memory.
What to ask for in the pack
- Physical sample per item, 20-50 grams in a sealed foil bag, with the packing or roasting date marked - not just the harvest year.
- A spec sheet: origin down to county or village where honest, cultivar or leaf variety, picking standard, season, grade, moisture claim if the packer tests it, and the format it will ship in.
- Cupping parameters: leaf-to-water ratio, water temperature and steep time the packer considers representative. If they cannot tell you, that is itself information.
- A reference sample kept by the supplier - the sealed counter-sample your bulk lot will be judged against.
- Documents: batch or lot record, and where applicable the lab panel (pesticide, heavy metal, microbiology) for the lot the sample came from.
Ask that samples come from the same batch you will receive where the program allows it, or at minimum from the same blend run. A sample pulled from a show-room jar tells you about the jar, not the lot.
How to cup it properly
Use a fixed protocol so results compare across samples and across weeks. The industry's standard sensory method - known in China as the five-factor evaluation under GB/T 23776 - examines dry leaf, liquor color, aroma, taste and the wet leaf after infusion, with set ratios such as 3 grams in 150 ml or 8.3 grams in 110 ml for set times. You do not need the full ceremony to buy well; you need the discipline. Same water, same vessel, same clock, same order, and at least two people tasting. Brew each sample twice: once at the packer's stated parameters, once at your own program's parameters, because the same tea can sing in a gaiwan and sulk in a 2-liter urn. Write one line per factor per sample while it is in front of you. From the desk side, every sample we send carries its cupping sheet on those five factors - the practice is described on the quality control page.
The documents that should travel with samples
| Item | What it tells you | When it matters most |
|---|---|---|
| Batch or lot record | Origin, processing date, and that the sample maps to a real production run | Every program |
| Lab panel (per batch) | Pesticide residues, heavy metals, microbiology against your market's limits | EU, US, Japan and any retail chain with its own standard |
| Organic certificates | The chain from certified material to your lot, not just a logo | Organic channels; claims must trace to certified material |
| Packing material spec | Food-contact grade of film, bags or cans; PLA claims backed by material reports | Teabag programs and compostability claims |
| Counter-sample confirmation | That both sides hold the same sealed reference for the bulk judgment | Every program that will reorder |
Documents and samples should reference each other by lot number. A lab report for a different batch than the one in the bag is a common and quiet failure of sample rounds.
Staging the round: first pass, second pass
Run samples in two passes rather than judging everything at once. The first pass is a broad filter: three to six candidates, same category, cupped on one protocol, scored quickly on the five factors to drop the clear non-starters. The second pass is a narrowing: two or three survivors, cupped again a few days later - palates reset, teas settled from transit - this time at your program's own brewing parameters, with colleagues in the room and notes written before any discussion. Two passes cost a few days and prevent the most common sampling error: choosing a winner on first impressions that fade by the second session. For custom blends, the same logic applies across iterations - agree with the desk what each revision changes, so version three is a step from version two rather than a restart.
Red flags in a sample round
Some answers should make you slow down. Refusal to state brewing parameters, or parameters that change between emails. Samples with no packing or roasting date. Spectacular stories with no village, no season and no lot number - origin romance is not a specification. A lab report that predates the sample or names another batch. Pressure to approve on the first infusion of a first sample. And the mirror-image failure on your side: judging green tea a week after arrival in summer heat, or judging a roasted oolong the day it comes off the fire, when it needs rest. Give high-roast lots a few weeks; give greens cool, dark storage and judge quickly instead.
From sample approval to bulk
Approval is a document, not a feeling: confirm the winning sample as the sealed counter-sample, log its spec sheet and lot record, and state that bulk will be produced to match, checked by cupping against that counter-sample before shipment. Stock lots typically ship 10-15 working days after confirmation; custom blends and new packing formats need longer, commonly 35-60 days where development or tooling is involved - typical reference ranges, confirmed in writing per order. Keep the counter-sample for the life of the program; it is what makes reorders comparable across harvests, a discipline the note on blend consistency builds on. For the testing side - which parameters, which markets - the testing parameters note gives the working panel. If you would rather have this round run for you, send the category and market through the quote form and the desk will assemble a documented sample pack.