Preferment Scaling (Poolish / Biga / Levain) – Getting Pre‑Doughs Under Control

This topic is part of the SG Systems Global regulatory & operations glossary.

Updated November 2025 •
Weighing & Dispensing, Dough Absorption, Dough Temperature, Mixer Load Management, Traceability
• Production, Tech, QA, NPD, Planning, CI

Preferment scaling is the controlled preparation, sizing and use of pre‑doughs such as poolish, biga and levain in industrial bakeries. It covers how much flour, water and yeast or culture go into each preferment, how big each batch should be relative to the final doughs, and how those pre‑fermented masses are tracked, aged and folded into production in a way that is repeatable, documented and safe.

Artisan bakers can sometimes “wing it” with a bucket of starter and a feel for readiness. A plant running tens of tonnes per shift cannot. If the flour, hydration, age or inclusion rate of a preferment is drifting, you will see it brutally in dough rheology, proof times, volume, flavour and scrap – and you will have no honest explanation for auditors or customers. Preferment scaling is about dragging this notoriously vague part of bread‑making into the same level of control as any other critical ingredient.

“If nobody can tell you exactly how much flour and water are tied up in today’s levain, you’re not running a controlled bakery – you’re running a science project at factory scale.”

1) What We Mean by Preferment Scaling

In industrial bakery terms, preferments are pre‑doughs made ahead of the main mix, typically combining a portion of flour, water and yeast or culture and allowing it to ferment for a defined time. Common types include:

• Poolish – equal parts flour and water (100% hydration), with a small yeast dose, fermented to a specific maturity.
• Biga – stiffer preferment (often 50–60% hydration), generally shorter fermentation and firmer handling.
• Levain / sourdough starter – culture‑driven preferment with mixed lactic/yeast flora, refreshed on a defined schedule.

Preferment scaling spans three layers:

• Composition – How much flour, water and yeast/culture go into the preferment itself.
• Batch sizing – How big to make each preferment batch relative to final dough requirements and holding capacity.
• Dosing into final dough – How much preferment mass goes into each final dough batch, and how that affects total flour, water, yeast and salt in the formula.

In a controlled environment, all three are defined in recipes and master data, not in someone’s head. Any change is treated like a formulation change, because that is exactly what it is.

2) Why Preferment Scaling Matters in Industrial Bakeries

Poolish, biga and levain are used because they deliver real benefits:

• Flavour and aroma: Organic acids, alcohols and aromatics accumulate in preferments, giving complexity you can’t get from straight‑dough processes.
• Dough structure and machinability: Properly scaled preferments can improve extensibility, tolerance and volume, making doughs friendlier to dividers and moulders.
• Shelf life and staling: Organic acids and enzyme activity can slow staling and improve keeping quality.
• Reduced yeast and improvers: Some functions of improvers and high yeast levels can be replaced by time and fermentation, which many customers and retailers like on labels.

But preferments also bring risk:

• Over‑ or under‑mature preferments change dough strength and flavour dramatically.
• Poor scaling can push final dough salt, yeast and absorption out of validated ranges.
• Weak traceability means you cannot say which pre‑dough lot went into which finished product – a recall headache waiting to happen.

So preferment scaling matters because it is how you get the upside – flavour, structure, label benefits – without uncontrolled variability, ugly CPV charts and endless “Friday night dough” mysteries nobody can explain.

3) Types of Preferment and Their Scaling Implications

Different preferments behave differently with respect to scaling and control:

• Poolish (100% hydration, high liquidity)
  • Easy to pump and dose by mass or volume.
  • Contributes a significant chunk of total water; tightly linked to absorption.
  • More temperature‑sensitive; ferments faster, narrower maturity window.
• Biga (stiff preferment)
  • Handled like a dough: tubs, troughs, chunkers.
  • Contributes more to dough strength than water; scaling errors hit structure hard.
  • Sampling for maturity is trickier; temperature gradients matter.
• Levain / sour starter
  • Culture composition and refresh ratio are as important as flour/water ratio.
  • Scaling is a moving target – inoculation and feed levels dictate final contributions.
  • Requires a defined “life cycle” and refresh schedule, not just a static recipe.

From a scaling perspective, the common denominators are:

• Every preferment must have a formal formula (flour, water, yeast/culture, additives).
• Every final dough must have a formal mapping of flour and water coming from the preferment vs from fresh ingredients.
• Every preferment batch must have a known mass and composition at the point of addition.

Skip any of those, and “we use 30% poolish” stops meaning anything concrete.

4) Preferment Formulas and Contribution to Final Dough

Preferments are not magic; they are just doughs. Their scaling is easiest to handle when you see them in two layers:

• Internal formula: Flour, water, yeast/culture and any salt, sugar or improvers inside the preferment itself.
• External contribution: How much of that preferment (as a % of final flour or dough) is added to the main mix.

For each product, you should be able to answer, with numbers:

• How much of the total flour is in the preferment vs in the final mix?
• How much of the total water is in the preferment vs added later?
• What proportion of total yeast comes from preferment vs final dough additions?
• Does the preferment contain salt, and if so, how much of total salt does it represent?

These answers are essential for:

• Correct absorption calculations (you must avoid counting flour or water twice or missing them entirely).
• Scaling scientific changes – you can’t halve poolish without knowing what that does to flour, water and yeast in the total formula.
• Label and spec control – for example if you claim “X% of the flour is pre‑fermented”.

In a digital environment, this mapping belongs in your recipe management system and MES, not a spreadsheet on someone’s desktop. If you are still manually re‑calculating contributions during changes, you are inviting silent errors.

5) Batch Sizing – How Big Should the Preferment Be?

Batch size is where theory meets scheduling. You want preferment batches that are:

• Large enough to cover several final doughs efficiently.
• Small and frequent enough to keep age within a narrow, validated window.
• Aligned with available tanks, troughs or IBCs and with mixer capacities.

Practical considerations:

• Usage horizon: How many hours should a preferment be used for once mature (for example, 2–4 hours for poolish; longer for some levains) before it is considered over‑mature?
• Line demand: How many kg of preferment per hour does the line need at nominal run‑rate and product mix?
• Tank / trough geometry: Partial fills may behave differently than full fills (temperature stratification, gas retention).
• Contingency: What happens if a line stops and preferment ages beyond limits? Is there a re‑blend or discard plan?

From there, you can derive sensible batch sizes and preparation frequencies: for example, “make 1 000 kg poolish every 3 hours for Line 1, feeding 3 doughs per batch”. These rules then go into planning and MES schedules. If preferment batches are sized “because the tank looks about full” or “because that’s how we’ve always done it”, you don’t have a robust design; you have folklore with stainless around it.

6) Time, Temperature and Maturity Windows

Scaling is not just about kilograms; it is about time and temperature. Preferments live or die on their fermentation profile. Key control points:

• Inoculation level: Yeast or culture dose sets how fast the preferment will develop at a given temperature.
• Holding temperature: Bulk temperature control (cooling or heating) defines how predictable that rate is.
• Maturity criteria: Objective signs of readiness – time, temperature, pH/TA, volume increase, rheology – linked to product performance.
• Usage window: How long after reaching “ready” the preferment can be used while still delivering validated dough behaviour.

In a controlled set‑up, preferment scaling and scheduling are based on this understanding. You don’t just say “mix poolish at 18:00 and use it on night shift”; you say “mix at 18:00, hold at X °C, ready at 02:00 ± 30 min, use between 02:00 and 04:00 only, then discard or re‑purpose with QA approval”.

Ignoring this and letting maturity drift is how you end up with final doughs that behave completely differently on the same nominal recipe, and with CPV charts that show ugly seasonal or shift‑based clusters nobody can explain without admitting the preferment is basically uncontrolled.

7) Weighing, Dispensing and Identification of Preferments

At the nuts‑and‑bolts level, preferment scaling relies on boring but critical basics:

• Guided weighing of ingredients: Flour, water, yeast/culture for the preferment are weighed or metered with the same rigour as any other mix, using calibrated devices and guided terminals.
• Preferment batch IDs: Each batch has a unique identifier, with date/time, shift, operator and target maturity time recorded.
• Clear labelling: Tanks, troughs and IBCs carrying preferments are labelled with ID, product families they are intended for, maturity window and any allergen information.
• Usage recording: When a portion of a preferment is added to a final dough, the MES/eBR records which batch ID, how much was used and which final dough ID it went into.

Without this, traceability is shot. If a culture goes wrong, or a poolish is fermented under non‑compliant conditions, you will not be able to say which products are affected, or prove to an assessor that your lot‑based risk assessment is sound. “We think it was yesterday’s poolish” is not going to cut it in a recall discussion.

8) Integration with Dough Absorption, Dough Temperature and Mixer Load

Preferments carry real flour and water. Ignoring that and treating preferment mass as a black‑box “ingredient” is how you corrupt your own maths. For each product, your systems should explicitly account for:

• Flour in the preferment as part of total flour – feeding into absorption, scaling, yield and legal‑weight calculations.
• Water in the preferment as part of total water – critical for both absorption and target dough temperature control.
• Mass contribution when calculating mixer load – preferment is part of the bowl weight that drives mixer torque and frictional heating.
• Inoculum and salt load – yeast, acids and salt coming from the preferment must be netted against final‑dough additions.

In practice, that means preferments are defined as full recipes in the master‑data system, and final dough recipes reference them as components with known flour/water/other content. If your formulation spreadsheets simply say “add 30 % poolish” without mapping the internals, your final dough specs, absorption and temperature calculations are already wrong on paper before you even touch the mixer.

9) Yield, Scrap and Cost Impacts of Preferment Scaling

Preferments cost money and can both help and hurt yield:

• Ingredient cost: Preferments “park” flour, water and yeast hours before use. Over‑production or poor scheduling leads straight to waste if batches expire.
• Line stability: Well‑scaled preferments can improve line speed and reduce scrap by giving more tolerant doughs; poorly scaled ones create sticky or weak doughs that shed scrap everywhere.
• Yield accounting: Flour and water consumed in preferments must be correctly attributed to the products they eventually feed. If your mass‑balance model treats them as a separate world, your yield variance analysis will be garbage.

From a finance perspective, preferment scaling should show up as:

• Clear yield models that include preferment consumption and waste.
• KPIs for preferment write‑off (volume discarded due to age or non‑conformance).
• Linkage between preferment process changes and changes in scrap, OEE and complaints.

If preferments are driving a lot of added labour, energy and write‑off without a clear product or yield benefit, that is a signal that your scaling and scheduling strategy is wrong. The answer is not to quietly cut corners on age or hygiene; it’s to redesign the system.

10) Roles & Responsibilities in Preferment Scaling

Because preferments touch recipe, microbiology, scheduling and line performance, fuzzy ownership is dangerous. A robust set‑up looks something like:

• NPD / Bakery development: Designs preferment types, compositions and inclusion levels for new products; defines target flavour, structure and shelf‑life benefits.
• Technical / Microbiology: Validates fermentation profiles, maturity windows, inoculum levels and food‑safety margins; owns change control for preferment formulas.
• Production (preferment area): Executes preferment recipes via guided weighing; controls sanitation, timings and tank/trough management; escalates deviations.
• Dough‑room / line operators: Use preferments according to MES instructions; do not “stretch” maturity windows or swap preferment types without authorisation.
• Planning: Schedules preferment preparation to match line demand; avoids creating queues of ageing pre‑dough “just in case”.
• QA: Audits records and practices; leads investigations when preferment issues appear in complaints or non‑conformances.

When any of these abdicate – for example, planning swings volumes wildly without discussing impact, or operators casually mix “similar” preferments between products – the whole control strategy collapses, often without immediate visibility until something goes badly wrong.

11) Common Failure Modes & Audit Findings

Preferments are a goldmine for findings when auditors look properly. Typical issues:

• No formal recipes: Poolish or levain made “by tradition”, with no controlled specification for flour, water, yeast/culture or hydration.
• Age and temperature uncontrolled: Vague statements like “ferment overnight” with no monitored time/temperature profile or maturity criteria.
• Weak identification and traceability: Troughs or tanks marked only with product name or date; no unique batch IDs; no link to final dough batch numbers.
• Unplanned use or reuse: Over‑aged preferments quietly blended into fresh ones or used on low‑risk products without risk assessment or documentation.
• Mismatched maths: Preferment flour and water not correctly mapped into final dough calculations, leading to inconsistencies in absorption, labelling or nutritional data.
• Data integrity gaps: Paper logs completed after the fact, times back‑filled, and recorded temperatures that don’t match actual conditions.

Any of these can be framed as process‑control and data‑integrity issues. If your story elsewhere is that you run a “highly controlled, validated process” and then auditors find preferments run on superstition, expect awkward questions about what “control” really means in your organisation.

12) Digital Preferment Management – MES, eBR and Historians

Modern MES and batch‑control systems can treat preferments as first‑class citizens rather than side projects. Typical digital features include:

• Dedicated preferment recipes with version control, linked to specific products or product families.
• Guided preparation workflows on weighing stations, enforcing correct ingredient selection, quantities and order of addition.
• Automated time/temperature logging from tank probes into a process historian, supporting maturity decisions and investigations.
• Electronic batch records tying preferment batches to final dough batches, including actual masses and usage timestamps.
• Alarms and holds when preferment age or temperature runs outside defined limits, or when the wrong preferment batch is selected for a product.

Done well, this turns preferment management from a black box into a transparent, auditable subsystem. It also gives CI and technical teams the data they need to optimise schedules, inoculation, temperature and inclusion levels based on what actually happens in the plant, not just in lab trials.

13) Designing a Preferment Scaling Strategy for a Site

Getting preferment scaling under real control usually involves more than just writing a few SOPs. A workable strategy typically includes:

• Product segmentation: Decide which products justify preferments (by volume, positioning, margin) and which should stay on simpler straight‑dough processes.
• Standardisation of types: Rationalise how many preferment types you run (for example, one poolish, one biga, one white levain, one rye levain) instead of letting every product have a bespoke pre‑dough.
• Validation of formulas and windows: For each preferment, validate composition, inoculum, time/temperature profile and usage window against product quality and safety.
• Scaling rules and scheduling templates: Define batch sizes, preparation frequencies and line allocation; bake these rules into planning and MES, not personal spreadsheets.
• KPIs and review: Track preferment‑related deviations, scrap, write‑offs and complaints; review them in technical and management meetings; adjust strategy based on evidence.

Expect some resistance from teams used to improvisation. The answer is not to crush all flexibility, but to channel it: capture what works in structured rules, and force any deviations to surface where they can be evaluated and, if appropriate, turned into new standards.

14) How Preferment Scaling Fits Across the Bakery Value Chain

R&D / NPD: Preferments are a design choice. Using them without considering industrial scaling, scheduling and hygiene is irresponsible. Development teams must own the consequences of “artisan” choices at plant scale.

Tech transfer: Scaling from a 10 kg test mixer to a 1 000 kg line means translating pre‑dough timing, temperature and inoculation rules into something planners and operators can actually execute. That includes backup plans for breakdowns and demand swings.

Routine operations: When preferment scaling is robust, lines see improved dough tolerance and flavour with predictable behaviour. When it isn’t, preferments become the excuse for every bad day and a permanent source of firefighting.

Supply chain and procurement: Flour and culture specs must support the preferment strategy. If procurement swaps flours, cultures or yeast types without looping in technical, your carefully tuned scaling can fall apart overnight.

Quality, regulatory and customer: Preferments influence everything from organoleptic profile to claims like “sourdough” or “long fermentation”. Being able to show customers and auditors exactly how you manage them – with numbers, records and risk assessments – is increasingly non‑optional if you want to play in premium or private‑label segments.

Across the chain, preferment scaling is either a quiet strength – a well‑understood, well‑behaved part of your control strategy – or a chronic source of unexplained variability. There isn’t much middle ground.

15) FAQ

Q1. Do we really need full recipes for poolish, biga and levain, or is “baker’s feel” enough?
In an industrial bakery, “feel” on its own is not enough. You can absolutely use experienced bakers to help set and refine preferment formulas, but those formulas still need to be written down, version‑controlled and executed consistently. Without formal recipes, you cannot demonstrate control, calculate real absorption or link flour and water usage to yield and label data in a defensible way.

Q2. How precise does preferment scaling need to be?
It needs to be precise enough that final dough behaviour and product quality are stable and that lot‑to‑lot variation is within your validated window. That usually means weighing or metering preferment ingredients and dosing preferment into final doughs by mass with tolerances similar to other major components. Casual scoop‑based dosing of “about one bucket of poolish” is not compatible with serious process control.

Q3. Can we blend old and new preferment batches to avoid waste?
Maybe – but only if risk‑assessed, validated and controlled. Some processes allow a defined proportion of mature preferment to seed a fresh batch or to be used in low‑risk products, but this must be codified in procedures, reflected in recipes and covered in HACCP. Ad‑hoc blending of over‑aged pre‑doughs to avoid write‑off is a great way to accumulate microbial and quality risk in ways you can’t see until they surface in complaints or shelf‑life failures.

Q4. How should preferments be included in dough absorption and temperature calculations?
Preferment flour and water should be treated as part of total flour and total water in the formula. That means mapping their content explicitly in your recipe system, then including that in absorption percentages and water‑temperature calculations. Ignoring the contribution or double‑counting it is a common source of quietly wrong formulations and unstable dough behaviour.

Q5. What are the quickest digital wins for controlling preferment scaling?
Fast gains usually come from three moves: (1) put preferment recipes into MES with guided weighing and batch‑ID generation; (2) link tank/trough temperature probes into a historian and display maturity status on operator HMIs; and (3) enforce selection of a valid preferment batch ID when starting a final dough, with automatic recording of the mass used in the eBR. These steps alone expose most of the hidden variability and “creative” practices around poolish, biga and levain.

Related Reading
• Dough & Fermentation:
Dough Absorption Control | Target Dough Temperature | Mixer Load Management
• Ingredients & Scaling:
Weighing & Dispensing | Minor & Micro Ingredient Stations | Absorption
• Quality, Data & Compliance:
Traceability | HACCP | MES | eBR