Take the sugar out of a biscuit or cake and you rarely get a less sweet version of the same product. It can come out smaller, darker or paler, firmer and with a different shelf life, because sucrose is a structural ingredient as well as a sweetener.[4][1]
This article explains what sucrose does in bakery, how bulking fibres, high-intensity sweeteners, polyols and allulose replace it and what each one costs you, and how to measure the result. It then works through an illustrative US biscuit brief using published trial data and sets out the FDA labelling rules that decide which claim a reformulated product can carry.
Sucrose is a large share of most sweet bakery recipes. In the pound-cake batter used in one 2026 study it was 25 g per 100 g of batter,[5] so removing it takes out mass, not just sweetness. A review of sugar substitution lists its roles: providing energy and sweetness, building structure, limiting starch swelling, giving colour and flavour, and extending shelf life.[2]
Because sucrose binds water, any replacement moves water activity (aw). In a 2026 pound-cake study, replacing sucrose with polydextrose raised aw from 0.816 to 0.871–0.887, with no significant difference between the 20%, 40% and 60% replacement levels.[7] When flour replaced sucrose in another pound-cake study, aw rose from 0.892 to 0.921 at a 37.6% reduction.[5] A higher aw in a moist cake is a shelf-life question that needs a real-time study; a dry biscuit starts from a much lower base, at aw of about 0.2.[8]
No single ingredient replaces sucrose. A practical system pairs a bulking agent for mass and water binding with, where needed, a high-intensity sweetener for sweetness, then re-tunes hydration and bake.[1][2]
Soluble fibres restore mass. In short-dough biscuits, replacing 30% of the sucrose with any of four soluble fibres gave biscuits whose sweetness on the trained panel did not differ significantly from the control.[6][8] Dextrin-type fibres behaved most like sucrose in the dough and during spreading, while the highest-polymerisation inulin-type fibre restricted expansion.[6] In pound cake, 20% polydextrose did not differ from the control in hardness (262 g against 358 g), but 40% and 60% did (656 g and 669 g).[7] For labelling, FDA allows several added fibres, including polydextrose and inulin-type fructans, to be counted as dietary fibre under enforcement discretion pending rulemaking, and uses 1 kcal/g for polydextrose.[9]
FDA lists steviol glycosides as 200–400 times sweeter than table sugar, sucralose about 600 times and monk fruit extract 100–250 times.[10] They are used at tiny levels, so they restore sweetness but none of the bulk, browning or water binding; they only work on top of a bulking system. For the health evidence on sugar replacers, see FRL’s review of the clinical evidence on sugar replacers.
Sugar alcohols are 25% to 100% as sweet as sugar, according to FDA.[10] They bring bulk but also label and tolerance limits: sorbitol is permitted at up to 30% in baked goods and baking mixes, and a food that could lead to a daily intake of 50 g of sorbitol must carry “Excess consumption may have a laxative effect.”[11] Allulose is treated differently on the US label: FDA exercises enforcement discretion to exclude it from Total Sugars and Added Sugars and uses 0.4 kcal/g for it.[12]
| System | What it restores | Main trade-off | Source |
|---|---|---|---|
| Resistant dextrins (soluble fibre) | Bulk; dough and spread close to sucrose | Firmer, crunchier bite; darker surface on the panel | [6][8] |
| Inulin-type fibres | Bulk | High-DP grades limit spread; one lower-DP grade darkened colour and raised bitter and burnt notes | [6][8] |
| Polydextrose | Bulk, water binding | Higher aw; darker crumb; harder cake at 40–60% | [7] |
| High-intensity sweeteners | Sweetness only | No bulk, browning or water binding | [10] |
| Sorbitol and other polyols | Bulk and some sweetness | Usage limit and laxative statement; sugar alcohol declaration when a sugar claim is made | [11][13] |
| Allulose | Bulk and sweetness | Excluded from Total and Added Sugars under FDA enforcement discretion, not yet by regulation | [12] |
Takeaway: choose the bulking agent first, because it decides spread, colour, aw and texture; the sweetener only corrects taste.
Each sugar function has an instrument, so a reformulation is judged on numbers as well as on taste.
| Method | What it measures | Published signal |
|---|---|---|
| Texture analysis | Hardness, fracture | Cake hardness 262 g (control) to 669 g at 60% polydextrose[7] |
| Colour (L*, ΔE*) | Browning | Crumb L* 71.85 to 65.61 at 40% polydextrose;[7] biscuit ΔE* about 4.0 with one inulin-type fibre[8] |
| Water activity | Water binding; shelf-life driver | Cake aw 0.816 to 0.871–0.887[7] |
| Dimensions | Spread, expansion | Biscuit width 35.80 to 34.27 mm with the highest-DP fibre[8] |
| Trained panel profile | Firmness, crunch, off-notes | All four fibre biscuits rated firmer and crunchier than the control[6] |
| Consumer test | Liking, purchase intent | Liking 6.36 to 4.19 at a 37.6% flour-only cut (62 consumers)[5] |
Takeaway: ΔE* alone can miss what a panel sees. In the biscuit study, every fibre biscuit was rated significantly darker on surface colour, including one with a measured ΔE* of about 0.9.[8]
Apply the science in a fixed order. Bake and measure a control, replace the bulk first, check size, colour, moisture and aw, and add sweetness only if the panel asks for it. Changing bulk, sweetness and bake together hides the cause of any failure (FRL practice).
The same gated approach is set out in FRL’s guide to the food formulation process from brief to stable prototype. Plan the calendar as well, including shelf-life work, with our food product development timeline.
Model-based design can shorten the loop. One 2025 study cut both sugar and fat in pound cake by 30% and, by tuning the water–sugar mixture to keep starch and protein setting temperatures and hygroscopicity on target, produced cakes that consumers rated comparable to the reference, with more sweetness, softness and moistness.[14]
This is an illustrative brief built from published data, not an FRL batch. The product is a short-dough biscuit sold in 30 g servings, which matches the FDA reference amount for cookies.[15] The control is assumed to contain 9.0 g of sugars per serving; the target is 6.3 g or less, which is 30% less.
Approach: replace 30% of the sucrose with a soluble fibre, hold a sweetener top-up in reserve, and check size, colour, aw, moisture and sensory against the control. The closest published system replaced 30% of sucrose with four soluble fibres in short-dough biscuits.[6][8]
What the published data predict:
Cutting sugar without replacing its function costs more. In pound cakes where flour replaced sucrose, reductions beyond 9.1% lowered preference, and at 37.6% purchase intent fell from 56.5% to 16.1%.[5] That makes bulk the first lever.
| Problem | Published signal | Lever (FRL practice) |
|---|---|---|
| Smaller biscuit | Highest-DP inulin-type fibre: width 34.27 vs 35.80 mm[8] | Use a dextrin or lower-DP fibre to keep spread |
| Colour shift | Lower-DP inulin-type fibre ΔE* about 4.0; others about 2.0 or less[8] | Screen fibres by ΔE* and panel; adjust bake |
| Firmer bite | All fibre biscuits firmer and crunchier[6] | Tune fat and hydration; blend fibres |
| Off-notes | Bitter and burnt notes higher with fibres[8] | Lower bake temperature or time; change fibre grade |
| Water activity drift (cakes) | Cake aw 0.816 to 0.871–0.887[7] | Re-measure aw each iteration; confirm shelf life |
Decide the claim before the formula, because each claim has different conditions.
The food must contain at least 25% less sugar per reference amount than an appropriate reference food. The label must name the reference food and state the percentage difference next to the most prominent claim, and must give the quantitative comparison per serving (for example, “lowered from 9 g to 6 g per serving”) next to the claim or the Nutrition Facts label.[3] In the worked example, 9.0 g to 6.3 g is a 30% reduction. Sugars are declared to the nearest gram on the Nutrition Facts label, so 6.3 g appears as 6 g.[13]
“Sugar free” needs less than 0.5 g of sugars per reference amount and per labelled serving, no sugar-containing ingredient unless flagged as adding a trivial amount, and either a low- or reduced-calorie qualification or a statement such as “not a reduced calorie food”.[3] “No added sugars” needs no sugars or sugar-containing ingredients added (including jam, jelly or concentrated fruit juice), no enzymatic increase in sugars, a comparable food that normally contains added sugars, and a statement that the food is not low or reduced calorie unless it is.[3] A wheat-flour biscuit cannot easily reach “sugar free”, so these claims usually need different formulas from a reduced-sugar product (FRL view).
The Nutrition Facts label declares Total Sugars and “Includes X g Added Sugars”, with a Daily Value of 50 g for added sugars. Where added sugars undergo non-enzymatic browning or fermentation, which applies to most baked goods, the manufacturer must keep records (for at least 2 years) to verify the declared amount.[13] Sugar alcohols are voluntary on the label, but become mandatory when a claim is made about sugar alcohols or total sugars, or about added sugars when sugar alcohols are present.[13] Allulose is excluded from Total Sugars and Added Sugars under FDA enforcement discretion.[12] Label errors after reformulation are a known recall trigger, as we explain in why reformulated food labels trigger US recalls.
FDA’s final rule redefining “healthy” was published on 27 December 2024, took effect on 28 April 2025 after a delay, and has a compliance date of 25 February 2028.[16][17] For a grain product, the food must contain at least ¾ oz-equivalent of whole grain, with added sugars no more than 10% of the Daily Value, sodium no more than 10% and saturated fat no more than 5%. Foods with a reference amount of 50 g or less, such as cookies, are assessed per 50 g.[16] The reduced biscuit in our example still carries about 10.5 g of sugars per 50 g. If most of that is added sugar, it is well above the 5 g (10% DV) ceiling.
In January 2025 FDA proposed a mandatory “Nutrition Info box” on the front of most packaged foods, rating saturated fat, sodium and added sugars as Low, Med or High. The proposal would also amend some nutrient content claim rules.[18] The comment period closed on 15 July 2025. We found no final rule in the Federal Register as of 27 October 2026, so treat it as a proposal and track it. For a label check against current rules, see our front-of-pack label audits and nutrient claims service note.
| Claim or rule | Key condition | Source |
|---|---|---|
| Reduced or less sugar | At least 25% less per reference amount; reference food, percentage and quantitative comparison declared | [3] |
| Sugar free | Less than 0.5 g per reference amount and per serving; calorie statement | [3] |
| No added sugars | No sugars or sugar-containing ingredients added; calorie statement | [3] |
| “Healthy” (grain product) | ¾ oz-eq whole grain; added sugars ≤10% DV; compliance 25 February 2028 | [16] |
| Front-of-package box | Proposed only (January 2025) | [18] |
The trial data come from short-dough biscuits and pound cakes; yeasted, laminated or filled products need their own tests (FRL view). Panels were modest: 25 panellists in the polydextrose study and 62 consumers in the flour-replacement study.[7][5] The biscuit colour differences quoted here were calculated by FRL from the published dataset means.[8] Dough rheology, hydration and bake profile can differ between bench and line, so recheck dimensions, colour, moisture and aw on the first pilot batches, write the acceptance table before the run, and confirm shelf life with a real-time shelf-life study.
Sugar reduction in bakery products succeeds when each job sugar does (bulk, water binding, browning, structure and sweetness) is replaced and then measured, not guessed. Check the result against the claim you plan to print, such as the US 25% reduced-sugar threshold, and treat the published figures here as a starting point for your own bench trials, since they come from other recipes.
If a biscuit, cookie or cake of yours must lose sugar for the US market without losing its bite, FRL can plan the replacement system and the bench trials around your recipe and target claim.
Start with FRL’s new product development service, add a shelf-life study of the final recipe, or contact the team to scope the work.
“Reduced sugar” needs at least 25% less sugar per reference amount than an appropriate reference food, with the reference food, percentage and quantitative comparison stated. “Sugar free” and “no added sugars” have separate conditions, including calorie statements.[3]
FDA exercises enforcement discretion to let manufacturers exclude allulose from Total Sugars and Added Sugars, and uses 0.4 kcal/g for it. This is guidance pending rulemaking, not yet a regulation.[12]
It can, because sugar binds water. In a pound-cake study, polydextrose replacement raised water activity from 0.816 to 0.871–0.887,[7] so shelf life must be re-validated with a real-time study.