Over the years I’ve had a few requests for soft serve help. Not from Mr. Softee competitors* or would-be usurpers of the throne of the Dairy Queen, but, maybe surprisingly, from chefs. They felt that something nostalgic (and maybe a bit kitschy) would work with their restaurant concept, but they wanted to keep their standards, stay in control, and make it all from scratch. Essentially, they’re working from David Chang’s playbook. But how do you do this with a traditionally industrial product like soft serve?
The hard part is that you need a soft serve machine. They’re expensive to buy, and probably not the most fun things in the world to maintain. But when they’re working, they do what they do, with very little user intervention. The difference between you, dear artisan ice cream maker, and the more typical soft-serve maker, is that you want to make your soft serve from scratch—and make well!—rather than buy some over-sweetened mystery and an artificial flavoring kit.
Is it Just Regular Ice Cream Made in a Soft Serve Machine?
Not quite. The fundamentals are similar, but the proportions are different. Here are some basic differences:
Serving temperature is warmer—typically -6°C to -8°C. Contrast this with -14°C for standard ice cream, or the softer Italian gelato styles that are served at -10°C to -11°C.
Fat is lower—typically 4% to 7%. Roughly half the fat of standard ice cream. For an richer than normal product, you can go as high as 10%. Higher than this, you’re likely to lose some of the lightness that people value in soft-serve. Higher than 12% (or lower than 4%) you can expect technical problems—either iciness or fat churning into butter. For reference, according to Douglas Goff, McDonalds uses 4% fat; Dairy Queen uses 5%.
Milk solids are higher. Generally, the lower the fat, the higher the milk solids. For a soft serve with 4% fat, you could go as high as 14% milk solids non-fat (MSNF). This is about keeping the total solids high enough to prevent iciness. I like to see total solids (fat plus non-fat solids) at 36%. The higher the solids, the more overrun you can achieve in the soft-serve while still maintaining satisfying body. Overrun can range from 30% to 60%.
Sugar is a bit lower. This is less about sweetness than about balancing the lactose in the high levels of milk solids. Typically 13% to 15%, compared with 15%-18% in standard ice creams. I usually push for lower sweetness than what’s standard, but this is accomplished through the blend of sugars. For soft-serve, I don’t drop the sweetness as low as I do for standard ice creams, because I don’t want to make them too “adult.”
Stabilizers are optional, and only used for texture enhancements. Typical usage is 0.2 to 0.3% total mix weight. I recommend using 0.15% the total water weight of the recipe as a starting point. You don’t have to worry about icy textures if your formula is properly designed and the machine is working properly (and you’re selling it reasonably quickly). So use a stabilizer blend that gives a texture that you like. [Correction 3-19-2023: earlier version recommended 0.35% by water weight. Typo]
Emulsifiers are required. You need them to get a proper dry surface consistency, good whipability, and good shape-holding qualities. Emulsifiers help control the foam structure, especially at the fat-air interface. Soft-serve is all about the foam structure. Typical usage is 0.1% to 0.2% the recipe weight. I recommend using 0.35% of the total fat weight as a starting point. A 1:5 ratio of Polysorbate-80 and Mono/Diglycerides works very well.
Flavorings can’t be chunky. You’ll gum up the machine. Use liquids, or solids that can dissolve or be completely dispersed in a high-powered blender.
Sample Recipe
Vanilla, with 10% fat—a relatively luxurious soft-serve
547g Whole Milk
230g Heavy Cream 36%
15g Vanilla Extract
60g Skim Milk Powder
112g Sucrose
10g Dextrose
20g Atomized Glucose DE40
0.5g Polysorbate-80
3.0g Mono/Diglycerides
3.5g Tara Gum (or 1.75g Locust Bean Gum + 1.75g Guar Gum)
0.8g Salt
1000g Total
Analysis
Total Fat: 10.1%
Milk Fat: 10.1%
Total Solids: 36.3%
Solids Nonfat: 26.2%
Milk Solids Nonfat: 11.4%
Acidity: 0.2%
Alcohol: 0.5%
Stabilizer/Water: 0.35%
Emulsifiers/Water: 0.35%
POD: 136 / 1000g
PAC: 263 / 1000g
Absolute PAC: 511 / 1000g
Rel. Hardness @ -8°C: 58 (for soft-serve the target is 55–60)
References
Goff, H. D., & Hartel, R. W. (2013). Ice Cream. 7th Ed. Springer Publishing.
*If you want to compete with Mr. Softee, bring a bat.
Hey underbelly, why dont you use your standard stabiliser mix of ratio 4:2:1 for locust bean gum:Guar gum:Lamda Carrageenan?
You could use that blend. But you don’t have to. You can keep it simple with something like tara gum (which promotes a rich texture) and not suffer any consequences. Because you don’t need the stabilizer for all the technical reasons you need it for in standard ice cream.
How do you feel about using only Tara gum in your base ice cream recipe instead of your normal mix of stabilizers? Would it be a 1:1 substitution? What are the downsides of Tara gum, since in the stabilizer section you made it sound like a nice combination of LBG and guar.
It works well, it just lacks the flexibility and fine-tunability of separate ingredients. You get the texture you get. If you like it, go for it.
Tara gum is a good stabilizer. Only downsides I see are 1) you have less control (you can’t vary the proportions to adjust the textural effects, as you can when using a blend) and 2) all the tara gums I’ve found hydrate at 82°c / 180°F which is a little hotter than my preference.
This is excellent thank you. Is the process just everything in the blender until smooth? Also, for chocolate base would it be simply the same swapping vanilla for cocoa powder?
It should be cooked / pasteurized just like standard ice cream, to denature the milk proteins. And you may need the heat to hydrate the stabilizers. Otherwise, yes—blend, chill, and pour in the machine (the machine may allow you to add hot ingredients; see the instructions).
You could swap cocoa for vanilla. You’ll get better results if you recalculate the formula to account for the added solids, fats, and bitterness in the cocoa.
But how can I tweak it to add flavours ether fruit-based or nut-based. I wish you could see this message thx
You can do anything as long as you balance the recipe to accommodate added fats, sugars, or water. And as long as everything is perfectly smooth. Fruits work if they’re well pureed (this requires reducing the other sugars, and increasing milk solids to compensate for the added water). Nuts might work best as infusions, rather than nut pastes blended into the mix. If you want to experiment with the latter, start small and increase gradually, so you don’t gum up the machine.
This is very helpful, thanks so much. But your emulsifier text says 0.35% of total fat weight, while in your recipe you’re using 0.35% of the total weight. Just want to check which is right, thank you
This is truly a very simple recipe but delivers great results. Thanks for sharing. If I wanted to turn this into a frozen yogurt would it be as simple as replacing the heavy cream and maybe adjusting the sucrose levels?
Probably not. But I haven’t explored this yet. Yogurts add viscosity but not whipable fat; I haven’t studied yet what makes them tick. If you want to experiment with this, I’d play around with a regular ice cream machine rather than risk gumming up a soft serve machine.
Any chance you can talk us through subbing egg yolks for the emulsifier? Also my taro powder gets clumpy no matter how I try to add it, what am I doing wrong? Thanks!
Can you use lecithin insted of the polysorbate and glycerides?
Yes, you should be able to. Glycerin is not as effective for creating a dry, airy, stable foam, but it should work. You may be trading quality for label-friendliness.
Any ideas on how to make a “drinkable” type soft-serve without a soft-serve machine? Coolish is a popular brand in japan (think one of those kid’s applesauce packets but soft-serve). It’s firm but not frozen at japanese 7-11 freezer temp, presumably -20°C, and can be extruded with reasonably few ice crystals.
From the ingredients list it looks reasonably similar to a typical soft-serve except for the oil which gives me pause, I can’t tell from just the ingredients whether it’s a cost-cutting measure vs. cream or if it has some functional property.
I’m thinking to lower the freezing point as much as possible, add extra emulsifiers (maybe glycerin for extra fpd and emulsification) and hope to incorporate a lot of air with my compressor machine — but I’d love to know whether this is completely crazy and I’ll just end up with an icy mess before I fully commit. If you have any idea what the dietary fiber and MCC are doing I’d also be interested, my best guesses are solids and gel respectively but I can’t think of much else that uses those two off the top of my head.
Ingredients:
Sugar, high fructose corn syrup, non fat milk, vegetable oils, egg yolk, fructose, dietary fiber, salt, emulsifier INS 471, stabilizer (INS 407 [mono and diglycerides], INS 410 [locust bean gum], INS 412 [guar gum], INS 415 [xanthan gum], INS 460(i)[microcrystalline cellulose]), artificial vanilla flavor, natural color INS 160b(ii) [annatto extract].
I’m not familiar with this kind of product. I think you’d want to serve it at a warmer temperature than typical ice cream serving temperature, rather than lowering the freezing point. Drinking something that’s -14C to -20C would probably mean instant brain freeze. I’ve never measured the temperature of a milkshake—that would be a good starting point.
What about a soft-serve sorbet mix? It’s really hard to find one. Do I omit the emulsifiers and just make sure there’s enough fiber in the recipe with either fruit selection or using inulin? I don’t know how the soft-serve shops in Montreal do it. Pregel and Itaberco soft-serve non dairy blends are like $32-$40 for 2.5 gallons excluding the cost of fruit. Impossibly expensive at standard prices and theoretically roughly 45 servings per batch.
It would be an interesting project. I’d be open to working on this (I’ll research it, but I bet there isn’t much good information out there). The only catch is that experimenting with soft-serve is expensive. The machines usually have big minimum batch sizes. There are some home models (like the Cuisinart) that are cheap and that might be good for development batches, but I have no experience with them.
Yes, from my experience, there is scant to practically zero information out there about this. I will buy from Itaberco until I can reverse engineer an acceptable product. I am no food scientist but I do have commercial machines and tremendous patience. My second lead is to Frankenstein your sorbet and softserve recipes. Else I will be forced to a Dole whip flavored breaking point.
I’m trying to figure out vegan soft serve and wow, there’s nothing out there. We just got a small machine for our restaurant and I figured we could use commercially available products and leisurely work on recipe dev, but the available products are not great. The first mix we tried was terrible, so sweet it was universally hated by everyone but a 9 year old. Then we tried a frozen vegan soft serve that was a lot better but has a weird off flavor.
I’ve worked with vegan cheese a lot but am new to ice cream and don’t even know where to start.
You’re right, there’s very little good information out there. Even for regular vegan hard ice cream, there’s nothing like the hundred years of accumulated scientific literature we have on dairy ice cream. I had to start with educated guesses and my own experiments.
You’re probably going to have to do quite a bit of experimenting to get something that works at all, to say nothing about getting something you love. I suggest getting your hands on one of the newer home soft-serve machines, so can run small test batches, and do so without risking damage to your commercial machine.
Send me an email if you’d like professional guidance on this. I can get you there faster, but it will still take some trial and error.