← All posts

Soap properties and formulation

Saponification tells you what quantities you need to manufacture a recipe. Property prediction tells you what you will actually get. This post covers the formulation analysis engine and how it derives predicted bar qualities from a recipe's oil composition.

What we built

The property prediction engine takes the oil ingredients in a recipe and calculates predicted properties of the finished bar: hardness, cleansing, conditioning, lather, longevity, iodine value, and INS value.

Each oil carries a fatty acid composition. The engine builds a weighted profile across all the oils in the recipe, proportional to their weights, then translates that profile into predicted soap properties using the formulation engine's property model. The results are stored against the recipe as calculated values, separate from the property definitions themselves.

The problem it solves

Soap properties are not attributes of individual oils. Olive oil is conditioning. Coconut oil contributes hardness and lather. Castor oil boosts bubble stability. But the bar you get is not simply a list of those individual contributions: it is the result of how those oils combine at their specific proportions.

A formulation tool that shows you properties per-oil does not tell you what the finished bar will be like. You need to model the composition of the recipe as a whole. An experienced soap maker develops intuition for this over time. The engine makes it calculable from the start.

The model also needs to leave room for other inputs to influence the result later. Additives, clays, and other ingredients can shift the base oil properties. The calculation architecture has to support that without the base calculation becoming entangled with it.

The challenge

The fundamental requirement is a whole-recipe calculation rather than a per-ingredient lookup. That part follows naturally from the fatty acid profile approach: each oil contributes to the overall composition in proportion to its weight, and the result reflects the recipe as a whole.

The harder problem was how to present those findings. The system is deliberately unopinionated. A bar that will be softer is not a bad recipe: it might be exactly the right shaving bar. A bar with high cleansing is not inherently good or bad either, depending on what it is for. The system reports the properties as calculated values. It does not flag deviations from an assumed ideal or push the soap maker toward a particular result.

Getting that right matters more than the calculation itself. A tool that second-guesses the soap maker's intent is a tool they will stop trusting.

What makes this different

The calculation is recipe-scoped, not ingredient-scoped. You are not looking up a hardness value for coconut oil and adding it to a hardness value for palm oil. You are calculating what a recipe made from those oils at those proportions will produce. That is a meaningful distinction when a formulation has eight oils and the ratios matter.

The results are stored as calculated values on the recipe, not recomputed on display. A soap maker can look at a recipe and see its property profile without the application re-running the calculation every time. When the recipe changes, the calculation can be updated explicitly.

What this unlocks

A soap maker can formulate a recipe in Plan.Make.Trace and see predicted bar properties before committing to a production batch. Adjusting the oil composition updates the profile. That feedback loop is the core of recipe development.

It is also the foundation for the additive property engine, which is in development. The additive engine is designed to take the next step: showing during recipe formulation what effect ingredients such as clays, sugars, sodium lactate, and other additives are expected to have on the finished bar. That will ship in an experimental mode after the initial beta.

What's next

The next post covers orders, suppliers and partnerships: how Plan.Make.Trace models the commercial relationships and purchasing activity that bring materials into the manufacturing process.

Liked this post? Tap the heart to let us know.

Get early access

Be the first to hear about early access, roadmap updates, and exclusive early adopter pricing.

Join the early access list