Scaling formulas up and down: yield, costing and the things that don't scale

A production bakery rewrites its formulas every day: 40 loaves on Tuesday, 220 for a Saturday market, a one-off 12 kg of croissant dough for a hotel order. Baker's percentage exists precisely so this is fast and error-free - but it only solves half the problem. Percentages scale the ingredients; they say nothing about how much dough to make for a target number of finished loaves, or about the handful of variables that stubbornly refuse to scale with batch size. This article covers all three: the scaling maths, yield (loss) conversion, and the non-linear adjustments. See img-scaling-flow-diagram for the whole method on one page.

For the underlying language of formulas, read the companion article Baker's percentage: the universal language of professional formulas (A8); for the production-process side of growing a bakery, see Scaling up: artisan to industrial (A5).

1. The scaling maths (recap, then the one formula you need)

In baker's percentage every ingredient is a percentage of the total flour weight, and flour is always 100% [c1]. To get a weight back from a percentage: ingredient weight = (ingredient % / 100) x flour weight [c2]; to get a percentage from a weight: (ingredient weight / flour weight) x 100 [c3].

The single most useful relationship for scaling is:

Total dough weight = flour weight x (total formula % / 100) - and, rearranged, flour weight = desired dough weight / (total % / 100) [c5].

King Arthur's professional shortcut packages this as the Formula Conversion Factor: add every percentage in the formula, then divide your desired dough weight by that sum; multiply each ingredient's percentage by the factor to get its weight [c4]. Worked on a lean French bread - flour 100% + water 65% + salt 2% + yeast 1.25% = 168.25% - a 150 lb batch gives a factor of 150 / 168.25 = 0.8915, so flour 89.15 lb, water 57.95 lb, salt 1.78 lb, yeast 1.11 lb (which sum back to 150 lb) [c4]. The same arithmetic runs in kilograms unchanged. This is the formula-lean-baguette-ddt card, and the worked table is in img-bakers-percent-worked.

Two ratios travel with the formula and are unchanged by batch size - which is the whole point: hydration = (water / flour) x 100 [c6], and water absorption, defined on flour weight (60% absorption = 60 parts water to 100 parts flour; cookie doughs sit at ~50-54%, artisan ciabatta up to ~90%) [c7]. A 1 kg test batch and a 100 kg trough at the same hydration behave the same way chemically.

2. Yield: how much DOUGH for how many FINISHED units

Here is the mistake that costs money: scaling to the finished weight you want to sell. Dough loses weight in the oven and during cooling, mostly as evaporated water - this is baking loss [c19]. You buy ingredients by dough weight but sell by baked weight, so you must gross the dough up.

Required dough weight = finished weight / (1 - loss fraction) [c22].

A 500 g loaf at 12% loss needs 500 / 0.88 = 568 g of dough; twenty 500 g loaves at 15% loss need (500 x 20) / 0.85 = 11,765 g [c22]. Typical losses: pan bread 10-12%, hearth bread and baguettes 12-15%, small rolls and buns 15-20%, croissants and laminated pastries 8-12%, cakes 5-10% [c19]. Smaller pieces lose proportionally more because they have more surface per unit volume - a ~50 g roll may lose ~18% while a ~1000 g boule loses ~12% [c20]. On top of baking loss, fermentation and scaling (dividing, handling) losses can add up to ~5% [c21], and higher hydration raises loss further (~12% at 65% hydration versus ~16% at 80%) [c24]. The reference chart is img-yield-loss-scale, with the divisors tabulated in the yield comparison table.

Two related production terms: dough yield (TA) = (total dough weight / flour weight) x 100 (net counts flour+water, gross counts everything, practical deducts losses) [c23]; and baked-goods yield, the quantity of finished goods from 100 parts flour, which depends on dough yield and baking loss together [c25].

Putting maths + yield together

The formula-white-pan-bread-scaled card is the vehicle. One pan-loaf formula (total 172.5%, hydration 62%) is scaled from a 1 kg bench test to a real order of 50 loaves at 800 g baked. The four steps: (1) scaling weight per loaf = 800 / 0.88 = 909 g; (2) total dough = 50 x 909 = 45,455 g; (3) flour = 45,455 / 1.725 = 26,350 g; (4) every ingredient = flour x its % - and they sum back to the dough weight. That is exactly the flow in img-scaling-flow-diagram. Weigh, don't measure by volume (img-digital-scale-dough-photo).

3. The things that do NOT scale linearly

This is where scaling goes wrong. img-linear-vs-nonlinear and the first comparison table summarise it; the essentials:

Dough temperature and friction. A larger batch generates more heat and takes longer to mix, so it leaves the mixer warmer. Control it with the desired dough temperature (DDT) formula - target ~24-26 C (75-78 F) for wheat yeast breads [c14]: water temp = (DDT x 3) - room temp - flour temp - friction factor [c15]. Worked as published by King Arthur: 78 F x 3 = 234; 234 - 72 - 71 - 22 = 69 F water [c16]. The friction factor is the heat your specific mixer adds; find it once by mixing a test batch and back- calculating - e.g. a batch reaching 26 C with room 22 C, flour 21 C and water 18 C has friction (26 x 3) - 22 - 21 - 18 = 17 C [c17]. Friction is roughly 22-24 F on a planetary/stand mixer and only 6-8 F by hand [c18] - and it rises with batch size and mix time, so the bigger batch needs colder water. With a preferment, multiply DDT by 4 and also subtract the preferment temperature [c15]. This calculation is img-ddt-formula-card and the formula-lean-baguette-ddt card.

Yeast form. Scaling often coincides with switching yeast type or supplier, and the forms are not interchangeable gram-for-gram because their water content differs: fresh ~70% water, active dry ~8%, instant ~3% [c11]. Convert by weight - fresh / 3 = instant, fresh / ~2 = active dry [c10] - so 100 g fresh is ~50 g active dry or ~33 g instant [c12]. In sweet/enriched doughs add ~5-15% more dry yeast (sugar slows yeast), or better, use an osmotolerant yeast [c13]. See img-yeast-conversion-chart. A practical food-safety note for scaling and stock: fresh yeast such as Lesaffre Benevia is perishable - store chilled 1-10 C (optimum 4 C), ~35-day life - while dried yeasts keep for months, which matters when you size an order [c40][c38].

Time and chemical leaveners. Mixing time grows with batch size; bake time tracks the size and shape of one unit, not the number of units - doubling a cake batch into a deeper tin needs a lower temperature and a longer (not doubled) bake [c27]. And chemical leavener does not scale safely into a deeper batter: baking powder sits at ~3% of flour in cake formulas [c9], but in a deeper tin too much gas rises then collapses with a bitter, metallic taste, so reduce it - at very large scale, leaveners and strong spices are commonly run at ~60-75% of the strict linear amount [c29]. When changing tins, scale batter by (new pan area / old pan area) at the same height and fill pans only ~1/2 to 2/3 [c28]. This is the explicit warning on the formula-butter-cake-true-pct card, which also passes the cake balance rules (sugar 100-160% of flour; fat 40-50%; egg >= fat; eggs + milk >= sugar) [c26]. Allergen review required before publication: the butter-cake card contains whole egg and milk — both Big 14 allergens (EU Regulation 1169/2011 / UK Food Information Regulations 2014); add a 'Contains: egg, milk' declaration to the formula card [c26].

4. Two more formula vehicles

A catalogue rye bread. The formula-mixed-rye-optimax card is a real production batch taken from the Zeelandia Optimax Free supplier specification, re-expressed in baker's percentage. It shows an improver dosed as a small fixed fraction of flour (0.1 kg per ~6 kg flour, ~1.7%) and a rye sourdough that contributes both flour and water - so the salt looks like 3.8% on added flour but is ~2.5% once the sourdough's own flour is counted [c36]. Process from the spec: mix 8 + 2 min, dough ~28 C, divide 0.58 kg into forms (~32 pieces from the 18.58 kg batch), final proof ~50 min, bake 250->230 C with steam for 5 min, ~45 min [c36].

A non-flour confectionery formula. Scaling logic is identical with a non-flour base at 100%. The formula-dark-ganache-base100 card uses dark couverture = 100% with cream as the variable: ~1:1 by weight for a pourable glaze (total 220%), ~2:1 for a firm filling/truffle [c41]. Ganache has no fermentation or baking loss, so ingredient scaling is purely linear - but larger masses cool and crystallise more slowly, so working temperatures and resting times must be adjusted at scale. Use couverture (EU: >=35% cocoa solids including >=31% cocoa butter) for clean melt and set [c42]; milk and white need less cream. Allergen and food-safety review required before publication: cream and optional butter are milk products (Big 14 allergen — add 'Contains: milk' declaration to the card); the 1:1 dark-ganache ratio is widely used in professional pastry but the Callebaut source URL returned 404 during verification — confirm against a current Callebaut technical guide before publishing [c41]; add a storage/shelf-life note for finished ganache (typically 2-3 weeks refrigerated for a 1:1 dark ganache).

5. Batch costing

Baker's percentage doubles as a costing tool. Cost a batch by summing (ingredient weight x cost per kg) for every line, then divide by the number of good finished units - not dough pieces - because you sell baked weight [c25]. Two scaling-specific cautions: include yield loss (you pay for 568 g of dough to sell a 500 g loaf [c22]), and remember an improver that already carries salt changes the recipe - IREKS Soft Roll 7 is dosed at 7% on flour and is 22.4% salt, contributing ~1.6% salt before you add any [c34], so the separately added salt (normally 1.8-2% [c8]) should be trimmed.

6. Catalogue: what to buy to run these formulas

The staple ingredients you scale, with first-party spec numbers, are in the key spec-sheet numbers comparison table. In short:

• Base flour (the 100%). Wheat Flour Type 550 (protein 11.5-12.5%, ash 0.51-0.58%, falling number

  =220 s [c30][c31]; img-flour-type550-bag), Domson White Strong Wheat Flour, Windrush Strong White; rye via Rye Flour Type 720 and Wholemeal Rye Type 2000; gluten boost via Hortimex Beneo VWG 75. Allergen review required: flour declares gluten; possible traces of soy, lupin and mustard [c31] — lupin can cause reactions in peanut-allergic individuals; confirm allergen statement against the current product label before publication, and note that the spec dates from 2025 (GoodMills Polska ed. 17).

• Yeast. Lesaffre Benevia and Lallemand fresh yeast (img-benevia-fresh-yeast) - mind the conversion factors and the chilled shelf life [c10][c40].
• Salt at ~2%. The Salt Company Extra Fine Sea Salt (NaCl ~99.9%, no allergens, vegan/kosher/halal/ coeliac [c32][c33]; img-salt-extrafine-bag).
• Improvers (dosed on flour). IREKS Soft Roll 7 at 7% [c34] (img-ireks-soft-roll-7), Zeelandia Optimax Free at ~1.7% [c36] (img-zeelandia-optimax-free).
• Cake + confectionery. Kent Foods Caster Sugar, Polmlek Unsalted Butter or Kruszwica cake margarine, Domson Baking Powder; couverture Callebaut 811 (54.5%), Barima Dark 72%, Zeelandia Arabesque 58%.

7. Faults and fixes

The scaling faults table lists the common failures - underweight loaves (loss ignored), over-fermented big batches (friction/water temp not recalculated), slack dough from a 1:1 yeast swap, collapsed cakes from flat leavener in deep tins, and costing that forgot yield. Read it before your next big batch.

All five formula cards in data.json show ingredient rows with percentages, a total percentage, a worked example at a real batch size, and process steps with temperatures and times; every percentage, conversion and yield figure in this article has been arithmetically checked for self-consistency.