The quality of a flavour is not created in a single moment of craftsmanship; it is built up from small decisions that check one another. The flow of an ordinary day explains this better than anything else.
In the flavour industry, quality is often a showcase word. Its equivalent on the ground is far duller and far more concrete: the sum of the control points a drum arriving at the gate in the morning passes through before it becomes a batch cleared for shipment in the evening. This article describes that day. There is no heroism in it; there is repetition, record-keeping and comparison.
Morning: not trusting an incoming batch on faith
The day starts with raw material intake. Every batch from a supplier arrives with a certificate of analysis. The certificate is valuable but not sufficient on its own; because a certificate describes what the batch looked like as it left the supplier's facility — not the transport, the temperature or the packaging integrity between there and here. That is why we repeat part of the control ourselves at intake.
We look for answers to three questions. Is this material genuinely the material it claims to be (identity)? Is there anything in it that should not be there (purity and profile)? And, at least as important: are its odour and taste the same as the reference we approved earlier? For identity and purity, intake control typically looks at physicochemical parameters such as refractive index and density; where necessary, the component profile is mapped with chromatographic methods and compared with the previously approved reference for that raw material. Is there a new peak, has a peak that should be there weakened — the answer sits right there.
With natural raw materials this comparison is especially critical. The compositional range of an essential oil shifts with the harvest year, the region, even the distillation time. The values on the certificate may stay within specification; the balance of components may nonetheless give a different character in the same formulation than it did last year. Compliant on paper, different on the nose. That possibility is what stops sensory control from being optional.
Quarantine: there is no such category as "probably fine"
A batch that fails intake control is placed in quarantine. For quarantine to work, two things are needed: the batch must be physically separated and the stock record must be closed to use. Otherwise quarantine remains a statement of intent. The physical separation matters, because on a busy production day the most likely error is the wrong drum standing next to the right one.
A quarantined batch has three ways out: released after additional analysis, returned to the supplier, or destroyed. Which route was taken and why goes on record. The real benefit here is not saving a single batch but accumulating supplier history. If a supplier's batches regularly arrive at the edge of specification, you would rather read that from the record than try to remember it three years later.
Weighing and blending: the most ordinary job, the most frequent error
When approved raw material goes to production, the work appears to get simpler: take the formulation, weigh, mix. Yet that formulation is the output of a long development process running from brief to series production; production's job is not to interpret it but to reproduce it exactly. In a flavour formulation a handful of components carry the main character; the rest are the details that make that character read as "real". A component mis-weighed at the one per cent level does not ruin the product outright — most of the time it produces a more insidious outcome: the product looks acceptable, it is simply "not what it used to be".
That is why several things are standard in weighing: a balance with current calibration, sequential component steps approved one by one, and volatile components added in a closed system with temperature under control. Nor are the parameters left free in blending.
- Mixing speed and duration — this is where the balance between homogeneity and volatile loss is struck.
- Process temperature — the upper limit for heat-sensitive top notes is specific to the formulation.
- Order of addition — some components will not enter the system unless pre-dissolved in a carrier.
- Equipment used and operator — the first thing to be read back if a deviation shows up later.
- Batch number and the raw material lots used — the lifeblood of the traceability chain.
All of these records are data nobody needs at the time. Their value emerges six months later, when a customer says "there is something different about that batch". A good answer to that sentence comes from the record, not from memory.
Afternoon: confronting the batch with its own reference
When the blend is complete the batch is not yet a product; it is a candidate product. The sample taken travels down two separate routes. The first is analytical: the batch's chromatographic profile is compared with the reference batch of that formulation. The aim is not to discover something new but to prove that nothing has changed. If there is a deviation, we discuss not only its magnitude but its direction — which component has weakened, which has come through more strongly than expected.
The second route is sensory, and it does not replace the first. The trained panel evaluates the sample blind, alongside the reference. Panel members score the notes individually; results are collected before any discussion, because the opinion of whoever speaks first shifts the perception of everyone at the table. The panel's question is simple but hard: can this sample be distinguished from the reference?
The critical point is this: a batch that comes out virtually identical to the reference analytically can still be distinguished on the panel. That is why both controls are mandatory; neither takes the place of the other.
“A batch is only as defensible as the record kept on the day; nobody can explain an undocumented batch retrospectively six months later.”
Evening: release, label and the trace left behind
If both controls are positive the batch is released. Labelling is the least discussed step and the one most open to error on the ground: batch number, production date, storage condition, application dose and which customer formulation the product belongs to. If the same flavour exists in two slightly different versions for two different customers, that label is the only thing separating them.
One of the industry's established practices is to set aside a retained sample from every batch and hold it under storage conditions. Most of the time that sample is never opened. The day it is opened is usually a bad day: there is a complaint at the customer and it has to be established whether the problem comes from the flavour, from a change in the customer's process, or from packaging interaction. Without a retained sample, that discussion rests not on evidence but on mutual assertion.
At the end of the day there is nothing spectacular to show. A series of controls has been repeated, a series of records kept, and traces left behind that will carry the answers to the next questions. That is the real lesson: in flavour, quality is a discipline rebuilt every day. The customer sees it not in the facility, but in consecutive batches coming out the same.
