Case study · Reverse engineering

Deformulating a benchmark body lotion: 98% of the formula decoded

A skincare brand in Africa asked CSL to find out what was really inside a body lotion it wanted to benchmark. CSL scientists combined analytical chemistry with data-driven modelling to turn an ingredient label into a quantified formula.

Body lotionAfricaAnalytical chemistryMachine learning
White pump-dispenser body lotion bottle with a green collar, the benchmark product CSL deformulated
Client and product names withheld under NDA
ClientSkincare brand, AfricaName withheld under NDA
ProductBody lotionBenchmark bought at retail
ServiceReverse engineeringDeformulation and benchmarking
Result98% decodedOf the formula, identified and quantified
01

The brief

What the client wanted to know

The client wanted to understand what was inside a competitor’s body lotion: not just the ingredients printed on the pack, but how much of each one the formula contains.

A full, quantified recipe would let the team benchmark the product, see where its cost and performance come from, and plan an improved version of its own.

02

The challenge

An ingredient label is only half the answer

A finished lotion is a multi-component system of water, emollients, emulsifiers, thickeners, humectants, actives, preservatives and fragrance. Turning it into a fully specified recipe means giving every one of those ingredients a percentage.

What the label tells you

  1. Which ingredients are presentMost, if not all, ingredients are declared on the label.
  2. That the amounts add up to 100%Every ingredient fraction together makes up the whole formula.
  3. The order of the amountsIngredients are listed from highest to lowest concentration. Those at 1% or less may appear in any order.

What the label does not give is the actual amount of each ingredient. Tens of thousands of cosmetic ingredients are available and only a handful end up in any one product, so a single label fits a very large number of possible recipes. Estimating the true concentrations is a hard, high-dimensional problem.

What the label gives

  1. Ingredient 1?%
  2. Ingredient 2?%
  3. Ingredient 3?%
  4. Ingredient 4?%
  5. Ingredient 5?%
  6. Ingredients ≤1%?%

What deformulation delivers

  1. Ingredient 1
  2. Ingredient 2
  3. Ingredient 3
  4. Ingredient 4
  5. Ingredient 5
  6. Ingredients ≤1%
Schematic only. Names and bar lengths are illustrative, not the client’s formula.
03

Our approach

Analytical chemistry, guided by data

CSL scientists are experienced in deformulation. For this lotion they combined laboratory analysis with machine learning where the problem demanded it.

  1. Step 1

    Decode the label

    Every declared ingredient was matched to its INCI identity, its likely function and its typical use range in body lotions.

  2. Step 2

    Analyse the product

    Analytical chemistry identified and measured the main components, using techniques from our reverse engineering panel such as GC-MS, HPLC, FT-IR, NMR and ICP-MS.

  3. Step 3

    Model the concentrations

    Where measurement alone could not fix a value, a machine learning model estimated it. The model was built on a curated formulations dataset from CAS, a division of the American Chemical Society, which records formulations in a consistent, highly structured way. Every estimate respected the label rules: all ingredients present, a total of 100% and the declared order.

  4. Step 4

    Reconcile and report

    Measured and modelled values were cross-checked and combined into one formula, with an estimated % w/w for each ingredient.

98%of the formula decoded

04 · The result

A label turned into a quantified recipe

CSL decoded 98% of the body lotion’s formula. The client received a quantified recipe it can benchmark against and build on, instead of an ingredient list with no amounts.

What the client received

Ingredient identificationEvery major ingredient identified, with its function in the formula
Concentration estimateAn estimated % w/w for each ingredient, from measurement and modelling
Analytical dataResults from each technique, with CSL scientists’ interpretation
Next-step optionsA route to an equivalent formula the client owns, with greener ingredient swaps
05

Why it matters

Deformulation for cosmetic scientists

Decoding a benchmark is a key capability for brands that want to stay up to date with competitors and develop new products to green chemistry standards with sustainable ingredients.

Stay ahead of competitors

Knowing what a leading product contains, and in what amounts, shows where its performance and cost come from.

Reformulate greener

A full recipe shows which ingredients can be replaced with sustainable, green chemistry alternatives without losing performance.

Develop faster

Starting from a decoded benchmark shortens the bench work needed to reach the texture and feel the market expects.

Related

Decode a benchmark of your own

Reverse engineering & benchmarkingIngredient identification, concentration ranges and an equivalent formula you own.FormulationCustom formulation and reformulation, built on the benchmark data.All case studiesMore projects from the CSL team, published with client permission.

Start a project

Create your next product with CSL.

Our formulation chemists, testing scientists and regulatory specialists work alongside your team, from the first brief to a market-ready product. Bring a brief, a benchmark or just an idea: a scientist will reply within one working day with next steps and an indicative scope.

Create with us