Plant-based milk foams like dairy, but only under a steam wand

Put a jug of plant-based milk under a steam wand and it behaves. Whisk the same carton with a battery frother and it either collapses or climbs higher than dairy ever does, with no way of telling which until it is in the cup. That is not a brand problem. It is a device problem, and a 2025 study put numbers on it.

Researchers at Ural State University of Economics foamed 10 plant drinks carrying Barista or Professional on the pack, alongside cow's milk, in three different ways, then measured how much foam came up and how much of it was still standing a quarter of an hour later. The drinks did not sort themselves into good ones and bad ones. The methods did.

Steam held every plant-based milk inside one band

Under steam injection at 65 °C the plant drinks produced foam volumes from 93% to 202% of the starting liquid. Cow's milk landed at 197%, inside that band and not at the top of it. The tallest steamed foam in the study came from an almond drink.

Switch to mechanical agitation at the same 65 °C and the band blows open. The plant drinks now run from 15% to 262%, with milk at 258%. One soy and almond blend managed 93% under steam and 15% when it was whisked. Same carton, same temperature, roughly a sixth of the foam.

The difference sits in how the bubbles get made. A wand injects hot vapour and shears the liquid at the same moment, which is why steam pressure matters more than hand technique for milk foam. A whisk only beats air in and hopes it stays.

Fat predicted the foam, protein did not

The expectation going in is protein. Protein is what holds foam together in dairy, so a plant drink with more of it ought to foam better. It did not work that way. Across the 10 drinks there was no correlation between foaming capacity and protein, and none with pH, density, total solids or total polyphenol content either.

Fat was the variable that moved. Whisked at 65 °C, fat content correlated negatively with foam volume at r = -0.81, p = 0.005, and at -0.64 when the same drinks were whisked cold. Under steam the same trend appeared but never cleared the usual significance bar, stopping at p = 0.08. The authors, writing in the journal Food Systems, put the missing protein signal down to the way these proteins interact with the stabilisers the drinks are built on, and they name gums and carrageenans as the next thing to test.

Cold is where the ranking flips

Whisked at 10 °C the spread runs from nothing at all to 196%. The coconut drink gave 0%, which the authors put down to its thick consistency and raised viscosity. Warm that same carton to 65 °C, the viscosity falls and it foams like the rest of them. An oat milk topped the cold group at 196%, just under dairy's 200%.

On volume alone, cold whisking looks survivable. Volume is not the measure that decides whether a drink still looks like anything by the time it reaches the table.

Fifteen minutes later the tall foams were gone

Foam stability was measured as the share of the original foam still standing 15 minutes after it was made. Steamed at 65 °C, the plant foams kept between 37% and 73% of themselves, against 45% for milk. The oat drink that had foamed so well in the cold kept the least of its steamed foam at 37%, and the soy and almond blend that had barely foamed at all kept the most at 73%.

Whisked hot, everything held better: 59% to 87% for the plant drinks and 65% for milk. Whisked cold, the plant foams kept 16% to 79%. The dairy foam kept none of it. A 200% head at the start and nothing left a quarter of an hour later, while an almond milk still had 79% standing.

What to change on the machine

Anything hot goes under the wand. If a machine has one and it gets skipped because plant drinks are supposedly awkward, that is the wrong way round, because steam is the treatment that closed the gap in the first place. Keep it clean while you are there, since milk residue is part of what turns a drip tray into its own small ecosystem.

A milk frother is not useless, but its range is narrower than the marketing around it suggests. For a hot cappuccino it can out-foam the wand, and that milk foam will hold longer too. Cold is where it runs out of road, and cold dairy is the worst case in the whole dataset.

So the next time a cold drink needs a head on it, whisk the plant carton and leave the milk in the fridge.

And the coffee under all that foam?

A milk drink is mostly milk, which is the reason the coffee underneath has to be loud enough to come through it. Santo Café's single-origin Mexican arabica carries enough body to hold its shape against a hot steamed foam. What lands in the cup still tastes of the coffee rather than of the milk.