From Pilot Batch to Full Production: What Changes When You Scale a Supplement

The formula works. The pilot came out clean: it tested true, it tasted right, it looked the part. So the full production run is just the pilot times ten, or times fifty: fill a bigger tank, run the line longer, ship more boxes. That is the assumption behind “just make more,” and it is the one that quietly puts launch dates at risk.

A supplement does not scale by multiplication. A bigger batch is a different physical environment, not a larger copy of a small one; the way it mixes, heats, holds, and fills all change with volume. And full production asks something a pilot never had to: not just that the batch comes out right once, but that it comes out right every run, on the same spec, for as long as you keep ordering. Those are two separate problems, and “make more” solves neither.

This is the jump from a pilot run to full production, from the floor: what actually changes when the volume goes up, what the pilot proves before you commit tooling and materials, the in-process checks that hold a product consistent from one run to the next, and how a small weakness the pilot would have caught turns into a failed Certificate of Analysis or a wave of complaints at scale. The earlier jump (from a bench sample to that first pilot) is its own subject, covered in how custom formulation works. Here we pick up at the next one, and at the harder question hiding behind it: holding the product steady run after run.

Short answer. Scaling a supplement is not “just make more.” A bigger batch mixes, heats, holds, and fills differently than a small one, so mixing behavior, fill weight, flavor, and dose uniformity all shift with volume. The pilot run (often in the hundreds of units when compatible materials are on hand) proves the formula survives production scale and speed before you commit components and line time. Full production adds a second demand the pilot doesn’t: the product has to come out identical every run, which is held by a locked master formula, in-process checks, and per-batch testing.

Best for: Founders and brands approaching a first real production run, and anyone whose bench or pilot sample worked but who hasn’t yet made it at volume.

Key decision: Whether to run the pilot that de-risks the scale-up, or commit tooling and materials to a full run on the assumption the formula translates unchanged.

Apollo path: Apollo Future Labs runs the pilot and the full production line from one floor in Livermore, California. A manufacturing quote returns the pilot size, the real component minimums, and the checks that hold your product consistent across runs.

Educational overview: not legal, regulatory, or medical advice. Requirements change and vary by jurisdiction and sales channel. Last reviewed July 2026.

“Just make more” isn’t scaling: the two things that change

Treat a production run as a bigger photocopy of the pilot and you inherit two problems at once. A scale-up that ignores either is where a launch slips: not because the formula was wrong, but because scaling was treated as arithmetic.

First, the physics changes again. A pilot proved the formula could be made on production-type equipment at a contained size. Going from that pilot to full commercial volume is another jump in scale, and the same variables that shifted on the way to the pilot keep shifting: a larger vessel mixes and transfers heat differently, a longer run holds product longer before the last unit is sealed, and a fill head running fast for hours behaves differently than a careful first fill. The formula does not automatically survive the second jump any more than it automatically survived the first.

Second (and this is the part a pilot never has to answer), full production has to repeat. A pilot is one batch. It has to come out right once. A product you actually sell has to come out right on run one, run two, and the reorder months later, each matching the spec your label and your Certificate of Analysis promise. Repeatability is not a bigger version of “make it work.” It is a different discipline: a locked formula, controlled equipment and process settings, checks during every run, and testing that confirms each batch against the same specification. A shop can nail a beautiful pilot and still make an inconsistent product if it has no system for the second problem.

Everything below is one of those two problems wearing different clothes: first the physics of a bigger batch, then the discipline that makes it come out the same way twice.

Why a bigger batch behaves differently: the mechanics the guides skip

Most scale-up write-ups will tell you “the physics changes at volume” and stop there. That is true and useless. Here is why it changes, because the mechanism is what tells you which formulas are risky to scale and which are not.

The root of most of it is simple geometry. As a vessel gets bigger, its volume grows faster than its surface area, so a large batch has proportionally less surface through which to gain or lose heat. It warms up and cools down more slowly, and it sits through a longer process from first mix to last fill. Mixing does not scale cleanly either: you cannot hold blade speed, mixing time, and energy delivered per unit of volume all constant when the tank grows, so something gives. Those two facts, slower heat exchange and non-proportional mixing, drive most of what follows.

None of these is exotic. They are the ordinary consequences of making more of something in bigger equipment over more time. The point is that they do not announce themselves: a formula with a marginal emulsion or a thin potency margin can pass a small, carefully made batch and only reveal the weakness once the process gets bigger and less gentle. That is exactly what the pilot exists to surface, and it is why “flavor drift” and “fill drift” are on the list of things that change, not things that stay put. When flavor is the sticking point, the deeper mechanics live in flavor masking; the takeaway here is only that a taste locked at the bench is re-checked at volume, never assumed.

The liquid problem the powder guides miss: fill behavior and big-tank mixing

Almost every scale-up guide on the internet is written for powders, capsules, and tablets. That leaves the hardest-to-scale format (liquids) barely covered. Liquids are where Apollo lives, from small vials to 55-gallon drums, and they scale on their own terms.

Three things make a liquid its own scale-up problem:

Big-tank mixing is not gentle. A stir bar in a beaker and an impeller in a production tank do not apply shear the same way, and the bigger the vessel, the more the flow pattern, the vortex, and the air pulled into the batch differ from the bench. An emulsion or suspension that came together by hand can break or settle in a larger tank, or it can need a different order of addition and a different mix time to hold. Higher-energy mixing also aerates: foam that a gentle bench stir never produced now sits on top of the batch, changing how it looks and how accurately it fills.

Hold time works against you. Fill a few hundred units and the batch is sealed quickly. Fill a full production run and the last unit is filled long after the first, so anything slow (a fine particle settling, a color shifting, two ingredients drifting apart) has real time to happen while the batch waits its turn at the fill head. A liquid that looked uniform in the tank can stratify before it is all packaged unless the process keeps it moving and the hold window is controlled.

The fill head is unforgiving. Hitting a small, exact volume repeatably (a 2-oz shot, a dropper dose) is its own discipline, and it is sensitive to the very properties formulation controls: viscosity, foam, temperature, and any suspended particles. A liquid that pipettes cleanly by hand can drift high or low moving through a fill head at line speed, and a fill that drifts is a product drifting off its label claim, unit by unit. A concentrated functional shot is the hardest version of this: a heavy active load packed into a small volume, where every one of these effects is amplified. It is also why a liquid-first floor tends to see fill and stability problems coming; it has run the format enough to know where a given liquid tends to misbehave on the way up.

Dry formats have their own scale-up traps (powder flow, blend segregation, compression and encapsulation drift), but the liquid version is where the SERP goes quiet, and it is the version most likely to surprise a brand scaling a shot, a tincture, a syrup, or a beverage.

What a pilot run proves before you commit tooling and materials

Here is the commercial reason the pilot matters, stated plainly: the pilot is cheap, and the full commit is not. Running the formula at a contained size (often in the hundreds of units when compatible materials are on hand) costs you a fraction of a full run. It is small enough to be affordable and large enough to be honest, because it is made on real or representative production equipment, the way the full batch will be made.

What you are protecting with that pilot is a real commitment on the other side of it:

• Materials. A full run means buying your components at their supplier minimums: custom labels, bottles, closures, and droppers that commonly land in the low thousands and that you own, whether or not this run uses them all. That is real money committed to real inventory. (The run size and those minimums are two different numbers with two different owners, which Component Minimums vs. Finished Runs pulls apart.)
• Tooling and line time. Any custom change-parts your specific container and fill require, plus a booked slot on a production line. Once that is committed, an off-spec batch does not just cost the batch; it costs the schedule.

The pilot is what you run before you sink either. Specifically, it proves:

• The formula still mixes, holds, and fills right at production scale and speed: the emulsion holds, the fill stays on target, the flavor and appearance survive the volume.
• The batch tests true after being made the way it will really be made, not the way a careful bench sample is made.
• The components fit the line (your bottle, closure, dropper, and label actually run clean at speed) before you buy them in bulk.
• The process is documented well enough to repeat, which is the whole subject of the next section.

A pilot does not erase risk (a formula built with no margin is a formula problem no pilot fixes), but it moves the discovery of problems to a few hundred units, where they are cheap, instead of to your first committed run, where they are not. If you are scoping one, how to prepare for a manufacturing quote covers what to bring so the pilot and the full run get quoted together, on one timeline, instead of as two disconnected asks.

The in-process checks that hold a product consistent across runs

Making one good batch is a formulation and process achievement. Making the same batch every time is a system. This is the half of scale-up the pilot introduces and full production lives on, and it is where vague guides say “we do in-process checks” without ever saying what they are. Here is what actually holds a product steady from run to run.

A locked master formula and batch record. When you sign off the pilot, the exact ingredients, grades, amounts, order of operations, equipment, and process settings become the master formula. Every production run is executed against it and captured in a batch record: the run’s own paperwork, showing what was used, what was checked, and who signed off. That document is the repeatability instrument: run two matches run one because both were made to the same written spec, not because the operator remembered. The detail lives in batch records and manufacturing documentation, but the principle is simple. If it is not written down, it is not repeatable.

Checks that bracket and sample the run, not just start it. A first-article check confirms the line is producing to spec before the run opens. Then fill weight, appearance, clarity, pH, and viscosity are checked at set points across the run (not once at the top), because drift is gradual, and a check only at the start misses a fill that slowly creeps low over hours. Retained samples are pulled and kept. None of this is optional theater: the FDA’s current good manufacturing practice rules for supplements (21 CFR Part 111) require a manufacturer to set specifications for a batch and confirm the batch meets them, with in-process points during production and a quality review before release. Consistency is a documented, verified outcome, not a hope.

Finished-product testing against the same spec, every time. Each production batch is verified against the same panels (identity, potency, and the contaminant and microbial limits your product requires), so a batch that drifted gets caught before it ships, not after. Which panels your specific product needs is its own decision, laid out in the testing your product actually needs.

One more consistency trigger that surprises brands: the manufacturer itself is a variable. Changing sites (moving to a new manufacturer, or splitting production across two) is a scale-up and a translation event of its own, because different equipment reproduces a process differently. It re-runs the same benchmark-to-pilot proof rather than starting cold, which is exactly what the manufacturer transfer playbook is built around. Years of runs, in cGMP-compliant operations on one floor, are worth naming here for a plain reason: a shop that has scaled many formulas knows where a given kind of product tends to break on the way up, and builds the checks to catch it before you do.

When a skipped pilot becomes a failed CoA

Skip the pilot and the physics does not go away; it just waits for your first committed run to show up, where a small formulation or process weakness stops being abstract and becomes a specific, expensive failure. The chain is short and worth seeing end to end.

• A thin potency margin plus fill drift. A formula sitting just above its label claim is fine at the bench, where you dose precisely. At line speed, a fill that drifts low tips finished units under their claim, and the finished-batch test comes back as a failed potency result on a full run of inventory you cannot ship. This is also where formulation overage, set from stability data, does its work: it is the deliberate margin that keeps a product on-claim through shelf life, and getting it wrong in either direction shows up as an off-spec CoA.
• A marginally stable emulsion plus hold time and a warm truck. An emulsion that holds for the day you taste it can break at scale, where the batch is mixed harder, held longer, and then shipped through heat. The result is separation, complaints, and returns on product that passed a same-day look but not a real-world one, which is why stability testing checks the product over time and conditions, and why a scale, process, or site change can re-open it.
• A low-dose active that segregates. An active that stayed evenly distributed in a small, gentle batch can separate by density or particle size at volume, and content-uniformity testing catches the uneven dose (again, on a full run).

The business end of every one of these is the same. A batch that fails its Certificate of Analysis cannot ship; a batch that shipped and then failed can become a recall; and on a marketplace, off-spec product and the complaints it generates drive returns and listing suppression. Learning to read a Certificate of Analysis tells you what “pass” and “fail” actually rest on, and the cheapest place to fail is a pilot, on a few hundred units, where a failure is a lesson instead of a loss.

Your pilot-to-production checklist

Before you commit a full run, these are the things the pilot should have proven. Use it to pressure-test any scale-up plan, Apollo’s included; if a manufacturer can’t tell you how each line is confirmed, that is your signal to keep asking.

The formula survived the scale-up

• It meets spec after production-scale mixing: identity, potency, and, for a liquid, clarity, viscosity, and stability.
• The emulsion or suspension holds; nothing breaks, settles, or separates on the timeline the real batch will sit.
• Flavor, color, and mouthfeel match the approved sample at volume, not just at the bench.

The fill is controlled

• Fill weight holds its target across the whole run, checked at points, not only at the start.
• The container, closure, dropper, and label run clean on the real line at speed, confirmed before you buy them at supplier minimums.

The batch is verifiable and repeatable

• It passes the same finished-product test panels your full run will have to pass.
• The process is captured as a master formula and batch record precise enough that a different operator, on a different day, gets the same result.
• Any settling, separation, or drift is checked against real hold and shelf conditions (the same sequence every run follows, laid out as the manufacturing map).

The commitment is scoped honestly

• You know the pilot size and the full-run component minimums as separate numbers, and which you own.
• The path from pilot to full production is one plan on one timeline, not two disconnected quotes, the way Apollo’s process is built to run it.

If a plan can check every box, “just make more” finally becomes true, because the make-more part is now the easy part, and the hard parts have already been proven.