Plant protein is bought on a number that does not mean what most specifications assume it means. “Protein 80 per cent minimum” appears on a pea protein specification, a soy concentrate specification and a sunflower protein specification, and in each case it is derived by a different route, expressed on a different basis and translated from nitrogen by a different factor. Two ingredients quoting the same figure can differ by several real percentage points of actual protein, and they will differ far more than that in how they behave in a formulation.
This guide is about the questions that sit behind the number.
The protein figure, decoded
Protein content in food ingredients is almost never measured directly. Nitrogen is measured, by Kjeldahl digestion or by Dumas combustion, and multiplied by a nitrogen-to-protein conversion factor. Three variables therefore sit inside every protein claim.
The conversion factor. The general factor of 6.25 is applied to most plant proteins in trade, but species-specific factors are lower for several materials and the choice can move the reported figure by several per cent. A specification that does not state the factor has not specified protein.
The moisture basis. “As is” and “dry basis” give different answers, and the gap is the moisture content of the powder. A dry-basis figure on an ingredient at eight per cent moisture is roughly eight per cent higher than the as-is figure for the identical material. Both are legitimate; using them interchangeably is not.
What the nitrogen came from. Kjeldahl and Dumas both measure total nitrogen, which includes non-protein nitrogen from free amino acids, nucleotides and other compounds. In most plant protein ingredients this is a small contribution. It is not zero, and it is larger in fermentation-derived and heavily hydrolysed materials.
So the first three questions to a supplier are: which method, which factor, which basis. A supplier who answers all three from memory is running a technical business. A supplier who has to ask the laboratory is reselling.
Flour, concentrate, isolate
The three tiers describe how much of the non-protein material has been removed, and each removal step changes cost, functionality and flavour.
| Tier | Typical route | Protein level | What else is present | Where it fits |
|---|---|---|---|---|
| Flour | Dehulling and milling of the seed or the defatted meal | Lowest of the three | Starch, fibre, minor lipids, most of the native flavour | Bakery, extrusion, cost-driven applications |
| Concentrate | Removal of soluble sugars by aqueous alcohol or acid leaching, or air classification for pulses | Intermediate | Insoluble fibre retained, most starch removed | Meat analogues, extruded texturates, nutrition bars |
| Isolate | Alkaline solubilisation, separation, isoelectric precipitation, neutralisation and drying | Highest | Very little; the process is designed to remove everything else | Beverages, dairy alternatives, high-protein applications |
Two process notes with commercial consequences.
Air classification is a dry route used mainly for pulses. It splits milled flour into a protein-rich and a starch-rich fraction by particle size and density. It is cheaper, it uses no water and no solvent, it preserves native protein functionality well, and it reaches concentrate rather than isolate levels. It also produces a starch co-product whose value determines whether the economics work at all.
Wet fractionation reaches isolate levels but denatures more protein, uses substantial water, produces an effluent stream and requires the pH swing to be well controlled. Excessive heat during drying is the most common cause of a functionally dead isolate that meets every analytical parameter on the specification.
Where a defatting step is involved, ask whether it is mechanical pressing or solvent extraction. Solvent-extracted meals give higher protein yields and are the industrial norm for soy; where hexane is used, residual solvent is a specification parameter and a question a buyer is entitled to ask. Mechanically pressed meals carry more residual oil, which shortens shelf life and can carry flavour.
Functionality is the property you are actually buying
Protein content tells you how much protein is present. It tells you nothing about what the protein will do. Four functional properties decide most applications.
Solubility is pH dependent and collapses near the isoelectric point of the protein. For most seed and pulse proteins that region is mildly acidic, which is precisely where many beverages sit. An isolate that dissolves perfectly in a neutral system can be unusable in a fruit-flavoured drink.
Gelation is the ability to form a heat-set network. It governs texture in meat analogues and in set desserts. It is sensitive to the thermal history of the ingredient, which is why two lots of the same product can behave differently after a dryer change.
Emulsification matters wherever fat and water have to stay together. This is also where lecithin and other emulsifiers enter the formulation as a complement rather than a substitute.
Water and oil binding determine yield and mouthfeel in formed products, and they interact with particle size: a finer grind hydrates faster and disperses better but is more prone to clumping and to dust.
None of these is captured by the protein figure, and most are not on a standard specification. Require functional data, on the specific lot type you will buy, and run an application trial before committing to a supply agreement. This is the single most common cause of a failed switch between nominally equivalent protein ingredients.
Flavour, colour and antinutrients
Plant proteins carry flavour from their source and from their processing. Pulse proteins carry a characteristic green, beany or earthy note that survives into the finished product; soy carries its own profile; sunflower and rapeseed proteins bring colour changes driven by phenolic compounds that darken under alkaline conditions. Masking these is a formulation cost, and it is one that a purely price-based comparison ignores.
Antinutritional factors deserve a specific mention because they are handled differently across sources. Trypsin inhibitors in soy are heat labile and are managed by the thermal treatment applied during processing; the level of that treatment is a real specification parameter for soy protein. Phytic acid binds minerals and is present across seeds and pulses; it is reduced but not eliminated by fractionation. Lectins are largely inactivated by heat. For most food applications at normal inclusion levels these are managed rather than problematic, but they belong in a technical discussion and not in a footnote.
Protein quality, and how it is expressed
Two systems appear in commercial documents. PDCAAS, the protein digestibility corrected amino acid score, was adopted following a FAO and WHO expert consultation and remains the basis for most regulatory and labelling frameworks. DIAAS, the digestible indispensable amino acid score, was recommended by FAO in 2013 as a technically superior successor because it uses ileal rather than faecal digestibility and does not truncate the score at one.
The practical points for a buyer: plant proteins are generally limited by one or two indispensable amino acids, commonly the sulphur-containing amino acids in pulse proteins and lysine in cereal proteins, and blending complementary sources raises the score of the mixture above either component. Do not compare a PDCAAS value from one supplier with a DIAAS value from another; they are not the same scale. What PDCAAS measures explains the calculation.
Documentation to require
- A full specification stating the protein method, conversion factor and basis, moisture, ash, fat, particle size distribution and the microbiological panel.
- Functional data for the properties relevant to your application, with the test conditions stated.
- Allergen status, covering the ingredient and the site. Soy is a listed allergen in the Union; pea is not. See allergen management in mixed facilities.
- GMO status for soy, with the documentation trail. Union rules on genetically modified food and feed and on traceability and labelling apply, and non-GM programmes rest on identity-preserved supply chains with segregation documented at every transfer.
- Solvent residue data where solvent extraction is used.
- Origin and sustainability documentation. For soy in particular, Union deforestation-free rules bring geolocation and due diligence obligations into the supply chain, and the evidence has to come from the origin.
- Lot-to-lot consistency data, ideally a control chart over recent production rather than a single certificate.
Comparing cost honestly
Price per kilogram of ingredient is the wrong comparison and it is the one almost every first-round negotiation uses. The right one is cost per kilogram of protein delivered into the finished product, which requires three adjustments.
Adjust for actual protein. Divide the price by the protein content on a common basis, using the same conversion factor for both candidates. An isolate at a higher price per kilogram is frequently cheaper per kilogram of protein than a concentrate.
Adjust for inclusion rate. If the lower-protein option requires a higher inclusion to hit the same declaration, it displaces other ingredients and changes the rest of the formulation cost, sometimes favourably and sometimes not.
Adjust for the formulation consequences. Masking flavour, adding a hydrocolloid to compensate for weaker gelation, or extending a process step to hydrate a coarser powder are all real costs attributable to the ingredient choice. So is a longer development cycle.
Only after those three adjustments does a price comparison mean anything, and it is common for the ranking to invert between the first and the third.
The five mistakes that recur
- Switching on specification equivalence alone. Two products meeting the same written specification can fail the same application. Trial first, always.
- Ignoring the pH of the finished system. Solubility data at neutral pH predicts nothing about performance in an acidic beverage.
- Treating a single certificate as evidence of consistency. Ask for data across lots.
- Leaving the allergen question to the label stage. It is a sourcing decision, not a labelling decision.
- Buying on protein content and discovering the flavour in the pilot plant. Sensory assessment belongs in the sample evaluation, not after the contract.
Soy against pea protein works the two dominant options through side by side, and seed oil cold pressing quality covers the co-product side of the same crush.