A useful microbiological requirement is a complete decision rule, not a line saying "microbiology: pass." For each test, identify the purpose, tea and intended use, target organism or indicator, sampling plan, analytical unit, method and edition, limit and units, point of application, and action for a non-conforming result. If any of those fields is missing, two laboratories can test the same lot correctly and still produce reports that cannot be compared.

Write the criterion before ordering the test

Criterion fieldBuyer should defineWhy it changes the decision
PurposeLot release, legal compliance, supplier verification, process trend or investigationA release criterion is not the same as a hygiene trend
Product and useTea type, lot, destination, repacking, direct brewing, beverage manufacture or validated later treatmentRisk and applicable rules depend on the real supply chain
TargetNamed pathogen, indicator, aerobic count, yeast or mouldOne result cannot stand for untested organisms
Sampling planNumber of sample units, analytical-unit mass, selection, pooling and acceptance numberTesting one portion does not define lot-level assurance
MethodStandard, edition, technique, matrix scope and validated alternative routeCounts and detection methods answer different questions
LimitPresence/absence or count, unit, rounding and treatment of less-than results"Not detected" and "below a count limit" are not interchangeable
Stage and actionWhen the criterion applies and who may hold, investigate, retest, divert, rework or rejectA result has no commercial effect until disposition is agreed

Dry does not mean sterile

Codex CXC 75-2015 for low-moisture foods explains that foodborne pathogens cannot multiply below the relevant water-activity conditions yet can remain viable for extended periods. Therefore low moisture does not prove absence of contamination. Dried tea is commonly managed as a low-moisture product, but the buyer should verify the actual product, water activity, process and intended use rather than infer safety from the word "dry."

End use matters. Tea sold for direct brewing, tea sent to a beverage factory and material entering a validated later treatment may need different verification. A future processing step changes a criterion only when that step and its performance are documented; a buyer should not assume that hot water, extraction or another process will correct an unidentified result.

Choose tests by the decision they support

Test typeUseful questionDo not conclude
Salmonella detectionWas the target detected in the specified analytical unit by the named method?A negative unit proves the entire lot is sterile
Aerobic colony countWhat count is recovered under the stated medium, temperature and incubation conditions?A low count proves absence of pathogens
Yeast and mould enumerationWhat viable count is recovered by a method applicable to the product's water activity?The test identifies every fungus, detects mycotoxins or predicts shelf life
Indicator organismDoes the chosen indicator provide useful hygiene or process information for this product?Any indicator substitutes automatically for a pathogen test

ISO 6579-1:2017, with Amendment 1:2020, is a current horizontal method for detecting Salmonella in food-chain samples. It is a method, not a universal tea limit or sampling plan. The U.S. FDA's current BAM Chapter 5 is the January 2026 edition. If a contract names BAM, record the chapter edition; "FDA method" alone is too vague.

For enumeration, ISO 4833-1:2013 plus Amendment 1:2022 covers colonies growing aerobically at 30 °C by the pour-plate technique. ISO is developing a replacement, so buyers should freeze the publication used for the order. ISO 21527-2:2008 remains published while revision is under development, but its own scope excludes dehydrated products at water activity 0.60 or below and it does not identify fungal flora or examine mycotoxins. Ask the laboratory to confirm method applicability to the submitted tea; do not select Part 2 merely because the product feels dry.

An eight-step buyer method

  1. Map the legal and customer layers. Check current destination law, importer controls and downstream specifications for the exact tea and use. The EU's Regulation (EC) No 2073/2005, current consolidated version dated 1 July 2026, sets criteria for specified food categories. Do not copy a limit from another category into tea; record the legal analysis and any national rule.
  2. State the decision purpose. Separate a pathogen release decision from indicator monitoring, supplier qualification, storage investigation or process trending. Codex says the purpose should be explicit.
  3. Describe the product and use. Record style, lot definition, moisture or water activity information where relevant, destination, consumer preparation, repacking and any documented downstream treatment.
  4. Select the target and method together. Ask the laboratory for the exact standard, edition, analytical unit, matrix applicability, detection or quantification limit and accreditation scope. Approve an alternative only through a defined equivalence or validation route.
  5. Design lot sampling before collection. Define sample-unit count and mass, random selection across the lot, individual versus pooled analysis, seals and chain of custody. Use the representative bulk-tea sampling method and obtain laboratory instructions before taking microbiology samples.
  6. Write acceptance syntax exactly. A presence/absence requirement must name the analytical-unit mass and acceptance number. A count requirement needs units such as CFU/g, limits, number of units and calculation rule. Never convert a supplier's generic target into law by description.
  7. Prewrite disposition. State who holds the lot, how the laboratory confirms presumptive findings, whether repeat testing is permitted, and the conditions for investigation, diversion, validated reprocessing, rejection or other action. Do not "test into compliance" through repeated samples.
  8. Trend separately from release. Store individual results with lot, date, method and qualifiers. A compliant lot can still contribute to an adverse trend; one adverse count may require process investigation even when a contractual acceptance number is not exceeded.

Read the report without losing its limits

"Not detected in 25 g" means the target was not detected in that analytical unit by the stated procedure. It does not mean zero organisms in a container or shipment. For counts, preserve the reported unit, dilution, quantification limit, rounding and any qualifier such as estimated count, less than the reporting limit or overgrown plate. Do not convert a less-than result to zero or compare CFU/g with log CFU/g without a documented calculation.

Compositing also needs control. Codex CXG 21 notes that pooling affects concentration and is not appropriate for enumeration methods or three-class plans; it may be considered for presence/absence testing only when performance is not compromised relative to testing individual units. The laboratory, criterion and contract should therefore agree on pooling before samples are combined.

End-product testing is evidence, not a replacement for preventive control. Codex says validated measures across the food chain offer more protection than sole reliance on acceptance sampling. Review the laboratory report alongside supplier controls, lot history, packing integrity and the COA evidence chain.

Keep five requirement types separate

  • Legal requirement: binding destination rules control market placement and cannot be waived by a sales contract.
  • Voluntary standard: ISO or Codex supplies a method or framework unless law or contract adopts it.
  • Trade reference: phrases such as "micro passed," "commercially sterile" or "standard plate count" do not define a criterion.
  • Buyer specification: the buyer selects a justified target, method, plan, limit and evidence route for the intended use.
  • Contract requirement: the signed agreement sets precedence, release authority, retest restrictions, remedies and cost allocation between the parties.

Common buyer mistakes

  • Copying numerical limits from a different product, market or customer without a legal and risk review.
  • Requesting "TPC" or "mould" without naming method, edition, units and analytical conditions.
  • Treating a low aerobic count as proof that Salmonella is absent.
  • Using a yeast-and-mould method without checking the tea's water activity against its scope.
  • Testing one convenient scoop and calling the result representative of the lot.
  • Pooling enumeration units or changing the regulatory sampling plan after collection.
  • Reading "not detected" as absolute zero or hiding reporting-limit qualifiers.
  • Repeating tests until one passes without a pre-agreed investigation and disposition rule.
  • Using end-product testing as a substitute for supplier hygiene and process controls.

Practical conclusion

The defensible sequence is define use - verify the legal and customer layer - state the purpose - select target and method - design sampling - write the exact limit - prewrite disposition - preserve and trend the evidence. This converts a vague test request into a criterion that a laboratory, supplier and purchasing team can execute consistently.

Add the completed fields to the bulk green tea purchase specification, align release evidence with Yunjing Tea's quality-control checkpoints, and send the destination, intended use and required test panel for a product-specific discussion.

Sources checked 13 August 2026: Codex CXC 75-2015, adopted 2015, revised 2016 and amended 2018 and CXG 21-1997, revised and renamed 2013; ISO 6579-1:2017/Amd 1:2020; ISO 4833-1:2013/Amd 1:2022; ISO 21527-2:2008; the FDA Bacteriological Analytical Manual and January 2026 Salmonella Chapter 5; and the EU's Regulation (EC) No 2073/2005, current consolidated version dated 1 July 2026.