Gelato Balancing Spreadsheet vs App: Excel's Real Limits

Almost every gelato maker starts with a spreadsheet. It is free, it is fast, and for the first twenty recipes it is genuinely enough. The interesting question is not whether Excel can balance a mix, because it can. The question is which parts of balancing it quietly refuses to do.

A balancing sheet is a model of your bench. The question is how much of the bench it can actually see.

What balancing actually asks a tool to do

Balancing is a constrained mass balance. You distribute 1000 g of mix across fat, milk solids-not-fat, sugars, other solids and water so that five or six derived numbers land inside accepted ranges at the same time. Change one ingredient and every derived number moves.

Those ranges are artisan practice, not law. The composition science behind them is set out in Marshall, Goff and Hartel, Ice Cream, and the US regulatory floor for the separate product called ice cream sits at 10% milkfat under 21 CFR 135.110, which is why American ice cream sheets and Italian gelato sheets do not use the same targets.

Two of the six rows are indices rather than measurements. POD and PAC are ratios against sucrose, and PAC in particular is not folklore. Freezing point depression is a colligative property, so at equal mass a sugar depresses the freezing point roughly in inverse proportion to its molecular weight. Sucrose has a molar mass of 342.3 g/mol; dextrose and fructose sit at 180.2 g/mol. Divide one by the other and you get 1.90, which is exactly why dextrose carries a PAC index near 190 when sucrose is set to 100. Lactose, also 342.3 g/mol, lands back at 100.

What a spreadsheet does genuinely well

Everything in the table above is a weighted sum. Each ingredient row carries percentage columns for fat, MSNF, sugar and water; each metric is one SUMPRODUCT away. A spreadsheet is an excellent tool for that shape of problem, and there are three jobs it does as well as any dedicated software.

Forward calculation. Given quantities, compute the metrics. This is linear, transparent and auditable. You can see every number and trace where it came from, which is more than can be said for a lot of black box calculators.

Costing. Cost per kilogram of mix, cost per served portion, margin at a given price. Spreadsheets were built for this and they remain the better tool.

Proportional scaling. Multiplying a validated 1000 g recipe up to a 6 kg batch is a single multiplication. A dedicated scaler does the same thing, but there is nothing here a sheet cannot handle.

Quick reference. A spreadsheet is excellent at the forward problem, meaning given quantities compute the metrics, and poor at the inverse problem, meaning given target metrics find the quantities.

Figure 1 - The forward path is a weighted sum any sheet can compute. The inverse path is an underdetermined system with constraints, which is where spreadsheets stop.

Where the spreadsheet quietly breaks

Nobody balances forwards. You do not sit down with 84 g of dextrose already chosen. You sit down with a target: 38% solids, PAC 26, POD 17, and you need to find the quantities that get you there. That is the inverse problem, and it is a different mathematical animal.

With six free ingredients and five target metrics, the system is underdetermined. There are infinitely many solutions, most of them useless because they violate constraints you never wrote down: dextrose above roughly 25% of total sugars starts to taste harsh and hygroscopic, glucose syrup has to stay high enough to control body, MSNF cannot climb past the lactose limit. Excel ships Solver, and Solver can technically do this. In practice it needs to be configured per recipe, it silently returns a local optimum, and when the targets are infeasible it reports failure without telling you which constraint was the blocker.

Four other failure modes show up long before that:

1. Ingredient drift. Each recipe tab holds its own copy of the sugar table. Your supplier changes the dry matter of a fruit paste and you now have forty tabs carrying a stale number, with no way to know which ones matter.
2. No guardrails. A sheet will happily hand you a mix whose lactose exceeds the crystallisation threshold. Goff and Hartel put that limit near 12.6 g of lactose per 100 g of water in the mix; above it you risk the gritty defect described in sandy gelato texture. Nothing in Excel warns you.
3. Fruit and Brix. Sorbetto balancing needs the sugar already inside the fruit, and that varies by lot. Feeding a refractometer reading into the sugar total means a lookup table plus a correction, and most homemade sheets skip it entirely.
4. One person, one file. The sheet lives on one laptop. Your second gelatiere cannot open it at the bench, and there is no history of who changed the fat target last March.

The freezing curve is the real dividing line

PAC is a linear proxy for something that is not linear. What actually determines whether a tub is scoopable at -13 C is the fraction of water frozen at that temperature, and that fraction is set by a feedback loop: as ice forms, the remaining unfrozen phase becomes more concentrated in sugar, which depresses its freezing point further, which limits how much more ice can form.

Solving that means iterating. You pick a temperature, estimate the ice fraction, recompute the concentration of the unfrozen serum, recompute the freezing point, and repeat until the numbers stop moving. Goff and Hartel describe this as the standard approach to constructing a freezing curve for a frozen dairy mix. A spreadsheet can do iteration with circular references enabled, but almost nobody sets that up, and the moment you do, the transparency that made the sheet attractive is gone.

This is why a single PAC number is a summary, not the whole picture. Two mixes with identical PAC can behave differently at the display case if their sugar spectra differ, because the ice curve between -18 C and -10 C is shaped by which sugars are present, not only by their aggregate index. If you want the reasoning in full, freezing point depression in gelato covers it.

When the spreadsheet is still the right answer

There is an honest case for staying in Excel, and it is not a small one.

If you run a fixed menu of twelve recipes that were validated years ago and rarely change, a sheet is fine. If your real bottleneck is costing and menu margin rather than formulation, a sheet is better than most apps. If you need a one-off calculation that no tool anticipated, nothing beats an empty cell. And a spreadsheet is portable in a way hosted software is not: the file is yours, it opens in ten years, and no subscription can revoke it.

The switch tends to make sense at three specific moments. When you start developing new flavours regularly rather than repeating a fixed set. When a second person needs to touch the formulas. And when you move into sorbetto, where fruit variability means the ingredient database has to be live rather than a frozen copy.

The useful framing is not spreadsheet versus app. It is which of the two problems you spend your week on. If you spend it computing, stay in Excel. If you spend it searching for quantities that satisfy competing constraints, you are asking a sheet to be a solver, and that is the one job it was never built for.

Related Concepts

• How to balance a gelato recipe walks the full procedure step by step.
• How to balance a sorbetto recipe covers the fruit side, where sheets struggle most.
• PAC calculator and POD calculator do the two index calculations directly.
• Total solids calculator checks the single most load-bearing number in the mix.
• Sugar selection guide explains why the sugar spectrum matters more than the sugar total.
• Gelato MSNF calculation guide shows where the lactose ceiling comes from.