From Trial Sample to Commercial Lot: Avoiding Scale-Up Surprises

A successful food sample shows that a particular ingredient worked in a particular trial. It does not establish that a larger batch will behave the same way in different equipment. Before moving a frozen ingredient into commercial production, record what made the small trial successful, identify the conditions that will change, and agree what the first production run must demonstrate. The useful comparison is between complete processes, from ingredient preparation through filling and storage.

For a buyer, that creates two connected decisions. Does the commercial ingredient meet the agreed supply specification? Does that ingredient perform acceptably in the customer’s production process? We recommend keeping separate evidence for each. A conforming delivery can expose a process limitation, while a successful recipe trial can hide an unrecorded change in the material. Combining both decisions into a single “sample approved” email makes later differences difficult to investigate.

A sample proves performance under recorded conditions

Start with the purpose of the original sample. An indicative bag may have helped the development team choose a fruit form. A lot-linked sample may have supported ingredient approval. A prepared filling may have demonstrated a flavour and texture target. None of those descriptions tells the production team how the material was mixed, heated, transferred or held. Record the intended decision before treating a sample as the basis for a commercial order.

For example, whole strawberries folded into a small pan may retain an appearance that the same fruit cannot retain through a proposed pump and narrow filling route. This is a question to investigate, not a reason to assume that whole fruit is unsuitable. If the finished product requires recognisable pieces, that requirement should influence equipment selection and trial observations before procurement commits to routine supply.

The ingredient specification still matters. Identify the product, form, relevant size or defect requirements, pack configuration and intended use. State whether the development material came from the proposed commercial supply route or was selected only as a target. If the source or preparation differs, keep the difference visible. An especially attractive hand-selected sample cannot silently stand in for the distribution of pieces in ordinary commercial packs.

A retained physical reference can help preserve the agreed appearance, provided its condition remains suitable. It cannot preserve the operating conditions that produced the finished sample. Keep the reference identity beside the trial record so reviewers can connect what they see with how it was made. The record should remain usable even when the original developer is unavailable to explain the missing details.

Capture the trial that worked

Reconstruct the successful trial while the team can still verify it. Use actual ingredient weights and lot identities, including added water and any liquid retained from the frozen ingredient. Record the formula version, batch size, equipment and working load. If an operator made an adjustment during the trial, record that adjustment as part of what happened. Copying only the planned recipe can erase the reason the sample succeeded.

Preparation deserves particular attention. Was the fruit added directly from frozen storage, partially thawed under a defined procedure, chopped, drained or blended first? Was the trial made from an entire small pack or from selected pieces? Did the developer add all the released liquid? These details can change the comparison with production. The trial record should describe actual handling without assuming that “frozen fruit” means one identical starting condition.

Describe addition order and mixing stages in operational language. “Mix until uniform” may be an observation, but it leaves unanswered who judged uniformity, where material was checked and what equipment was running. Pair the observation with the recorded settings, elapsed stage time and relevant product condition. Production can then decide which observations and measurements are suitable for the larger equipment.

Keep a distinction between measured values and recollection. A saved temperature trace is different from a remembered pan temperature. A weighed filled output is different from an estimated yield. Where a consequential condition was never recorded, identify it as unknown and consider a repeat bench trial. Filling gaps with plausible numbers creates an apparently precise comparison that has no reliable starting point.

Finally, describe success in the form in which it was assessed. A warm spoonful, a cooled cup and a filling baked into the finished product may answer different application questions. Name the assessment stage, preparation, time after processing and acceptance criteria. The production trial should reproduce the relevant evaluation, even when it cannot reproduce the small equipment.

Transfer the process, not just the ingredient percentages

Increasing ingredient weights does not automatically reproduce mixing, heating or material movement. The production team needs to translate the process using the actual equipment and product behaviour. A mixer speed written on a bench sheet is not enough to establish equivalent treatment in another mixer. Geometry, working load and the product’s path through the equipment also need review.

A useful published example comes from RSSL’s vegan cheese scale-up case. Its team found texture differences during pilot and factory trials. Investigations considered moisture and then the force applied by the factory mixer; further trials with reduced mixing force addressed the softer texture. This is a cheese case, not a prediction for frozen fruit. Its value here is the demonstrated need to investigate process differences instead of assuming that a matched formula guarantees a matched structure.

Use a short transfer table to expose the differences before the line is booked. The example below is a proposed handover structure. Fill it with your own records and agreed evaluation methods; it contains no recommended operating settings.

Prioritise differences that could affect the product’s defining attributes. A smooth sauce may tolerate a loss of visible piece identity that would fail a fruit topping. A ready meal may require vegetables to remain recognisable after the customer’s final heating step. The same incoming ingredient can therefore require different trial observations in different applications.

Do not turn this table into an equipment wish list. For every important difference, name the observation or test that will answer the resulting question. If nobody can explain what a measurement will change about the decision, reconsider whether it belongs in the trial. If an important decision has no supporting observation, add one before the run.

Follow frozen material through the actual line

The relevant process begins before the mixer starts. Observe how commercial packs are removed from storage, staged, opened, weighed and emptied. A small sample may be ready to add immediately, while a production quantity passes through several handling steps. Record the actual condition of the first and last additions rather than assuming that all material entered in the same state.

The frozen ingredient also becomes part of the batch’s thermal load. The production team should assess the temperature history resulting from its actual quantity, condition and addition sequence. A bench heating time should not be multiplied by the batch-size ratio or copied as an industrial endpoint. Use the facility’s defined process requirements and appropriate measurements to evaluate what the larger system actually does.

Follow pieces beyond the kettle. Material may wait in a holding vessel, circulate through a pump or travel through pipework before reaching the filler. Where piece integrity matters, compare suitable observations before and after a suspected damaging stage. Where distribution matters, examine filled product. An acceptable spoonful from the top of the kettle does not establish that every part of the filling run has the intended fruit-to-sauce balance.

Record interruptions as well as smooth running. An unplanned delay can expose a holding condition that a short development trial never encountered. Describe the event, affected material, process conditions and decision taken. Do not discard inconvenient observations merely because the line eventually produced an attractive sample. They may identify the operating limits that routine production needs.

Quality observations and food-safety validation have different jobs. FDA’s preventive-controls summary describes monitoring and verification, including scientific validation of process preventive controls for covered facilities. A taste panel or attractive finished pack does not provide that evidence. The customer’s qualified technical team must assess the applicable hazards and process requirements; this handover table does not establish a safe heating, cooling or storage procedure.

Give the first commercial trial clear release points

A commercial trial needs an agreed purpose, available material and time for decisions. Confirm whether the run is intended to explore settings, confirm an already developed process or provide evidence for routine production approval. An exploratory run may reveal useful limitations without producing releasable goods. Its commercial terms should reflect that possibility before the ingredient is consumed.

Set release points around decisions that could otherwise become assumptions. The sequence below is a working proposal. The sampling design, limits and authority must fit the product, facility and agreement.

1. Approve the trial input. Confirm ingredient identity, agreed specification, required documents, delivered condition and pack format. Identify any difference from the development material.
2. Authorise the planned run. The manufacturing team confirms equipment, working load, process instructions, checks, sampling locations and the response to a deviation.
3. Review the filled output. Assess the relevant physical, sensory and analytical criteria using identified samples. Keep evidence from appropriate points across the run, including interruptions where relevant.
4. Decide the material’s disposition. Record who may release, hold, investigate, rework or reject the affected output under the facility’s procedures and commercial agreement.
5. Approve the next production basis. Close significant deviations and define the formula, ingredient, equipment and operating conditions covered by the approval.

Start, middle and end samples can be a useful way to discuss variation during filling, but those three labels are not a statistically sufficient sampling plan by themselves. The team must choose numbers, locations and tests for the actual risk and variability. Keep sample identities tied to production times or events so an unusual result can be traced to the corresponding process history.

Include the final pack and relevant subsequent handling. RSSL’s chilled-dessert commercialisation case, for example, describes storage testing of samples in final packaging after pilot work. A favourable immediate trial result should not be stretched into a shelf-life claim. Define separately which storage or application work remains necessary and whether the current approval is conditional on it.

Make the purchasing decision equally explicit. Agreement that the ingredient met its delivery specification does not automatically allocate the cost of a failed line experiment to its supplier. Conversely, calling an order a trial does not excuse an ingredient that misses agreed requirements. Record quantity, permitted use and the route for investigating differences alongside the technical plan.

Diagnose a difference before changing the ingredient

Suppose a fictional peach filling trial shows fewer recognisable slices in filled packs than the approved development sample. The first task is to locate when that difference appeared. Check incoming form, preparation, mixing, holding and transfer in a controlled sequence. A photograph of the delivered slices can support an incoming observation, but it cannot establish what happened throughout the lot or inside the customer’s process.

If suitable incoming observations show the expected slices and a matched bench preparation still performs acceptably, investigate the line stages where breakage could arise. If incoming material itself differs against the agreed test, investigate that supply difference. These are working diagnostic branches, not proof that one party is responsible. Preserve samples and records that allow both explanations to be tested.

Change as few consequential variables as practical during the investigation. If fruit form, mixer setting, addition order and holding time all change together, an improved result does not reveal which change mattered. Sometimes several changes are necessary for safety or practicality; record them and acknowledge the resulting limit on interpretation. A controlled follow-up may still be needed before the team can define a repeatable process.

Compare yield on the same basis. Separate acceptable filled product from material left in the system, trial samples, rejected packs and planned losses. Record water additions and any draining steps so a larger gross output is not mistaken for improved usable yield. The useful commercial question is whether the agreed finished product can be produced consistently at an acceptable cost, not whether one trial produced the heaviest total output.

When a change appears promising, check the attributes it could affect as well as the one it improves. Gentler handling might preserve pieces while leaving distribution less uniform. A revised sequence might improve filling but change the product’s temperature history. The responsible technical team should decide what further quality and safety work is needed before the revised process becomes the approved production instruction.

Build a handover that operators can repeat

The final handover should state what was actually demonstrated and under which conditions. Include the ingredient and formula versions, relevant lot identities, equipment and working load, preparation and addition sequence, recorded operating conditions, sample results, yield basis and resolved deviations. Attach the approved records rather than leaving essential detail scattered across photographs and email chains.

Name the scope of approval. Acceptance for one product on one line does not automatically cover another line, a different pack or a wider operating range. If a condition remains open, describe the restriction and the person responsible for closing it. A conditional approval can support sensible progress when its conditions are clear; an unexplained “approved” can conceal unfinished work.

Give operators a usable instruction with defined checks and a response when the process moves outside its approved basis. Give procurement the corresponding ingredient description and change-notification requirements. CFIA’s supplier assurance guidance includes specification agreements, notification of changes and ongoing verification. Its Canadian guidance supports the value of keeping the supply agreement connected to the material that the customer has approved.

Before repeating the run, confirm that the approved documents reached the people who will use them. The objective is a commercial process that another qualified shift can reproduce within its defined conditions. A successful sample remains valuable evidence, but a complete handover is what makes that evidence usable for the next batch.

References

• RSSL: Vegan cheese scale-up case — an actual investigation of process-related texture differences.
• RSSL: Chilled-dessert commercialisation case — development, co-manufacturer trials and final-pack storage testing.
• FDA: Preventive Controls for Human Food — the role of monitoring, verification and process-control validation for covered facilities.
• CFIA: Supplier Food Safety Assurance Program — specifications, supplier agreements, changes and verification.