American Coffee is Too Sweet

Flavored coffee drinks – like pumpkin spice lattes – are at least as sweet as pastries, and that’s disappointing.

I love dessert. I’ll always ask for a dessert menu at a restaurant, and I’m disappointed when the only options are coffee and a scoop of ice cream. A bakery crawl is my idea of a solid weekend plan. My freezer is always – always – stocked with cookie dough and ice cream.

And yet, whenever I get a sweetened drink from a coffee shop, I almost always think that it’s far too sweet – cloyingly sweet, distractingly sweet, off-puttingly sweet. This mystifies me. I rarely shy away from a coffee shop’s pastries, but as a rule I avoid buying their eggnog cappuccinos, caramel macchiatos, and chai lattes.

What explains this apparent mismatch? Are coffee shops putting more sugar in their drinks than in their muffins? Or, is it possible that I perceive sweetness differently in solids and liquids?

How sweet are coffee shops’ drinks?

The simplest question here is: do coffee drinks typically have more sugar than baked goods?

Of course, there’s a lot of variation around the “typical” drink or pastry – are we talking about baked goods from my own kitchen, from my local bakery, or from Dunkin’ Donuts? Given this variation, I looked at the sugar levels in the foods and drinks from an establishment that’s large enough to post their nutrition information online, and that’s been taste-tested with a lot of Americans: Starbucks. I looked at their online nutrition information for 22 baked goods and 14 hot, sweetened drinks (i.e., black coffee isn’t part of this calculus) (see Table 1)1.

On average, each 8-oz Starbucks drink contains 17 grams of sugar (standard deviation = 4.5g). For reference, 8 oz of classic Coke contains 26 grams of sugar. 17 grams is less criminally sweet than I had expected, but is still a lot of sugar – depending on your source, 17 grams of sugar is roughly half of what you should consume each day2.

Are the drinks sweeter than the pastries?

In comparison, the average Starbucks baked good contains 24 grams of sugar (standard deviation = 9.8g). As the violin plot below illustrates, there’s also more variation among the pastries than the drinks. So if you’re sitting down to a full brownie or muffin from Starbucks, you’ll likely consume more sugar than if you sit down to a full 8 oz drink, though this will not always be the case.

Sugar per serving in coffee drinks and pastries from Starbuck’s. A “serving” is 8 oz. for drinks or a whole baked good for pastries.

Of course, the 8 oz. drink is the smallest size that Starbuck’s sells – if you got a 24 oz. drink, you’d consume three times as much sugar (an average of 51 grams) and outstrip the brownie-eater.

One bite at a time

But really my question is about the perception of sweetness as you’re chewing the food or swallowing the drink, not after you’ve finished it. To understand that aspect of the sweetness experience, it’s more informative to compare how much sugar is in a single mouthful of a Starbucks drink or a baked good.

To do so, we’d want to know how large a mouthful of a drink or a baked good is, and then calculate how much sugar that mouthful contains. Starbucks does not include this level of detail in their nutrition facts, so I had to do some field research. When drinking, I found that I swallowed about 1 fluid ounce (31.6 g. – about 2 tablespoons) at a time. Each mouthful of the average Starbucks drink would therefore contain about 2.1 g. of sugar. It’s harder to estimate how large my mouthfuls of baked goods are, because their densities vary a lot – a mouthful of a fudgy brownie probably weighs more than a mouthful of a fluffy muffin, for example. So, I measured my average mouthful sizes across a few baked goods of different densities: scones, muffins, cookies, and brownies. Averaging across these, I found that the typical bite weighed around 7 grams.

That means that each mouthful of the average Starbucks baked good contains …2.1 grams of sugar (sd = 0.65). This was not what I expected! My methodology was definitely imprecise. But it at least suggests that, mouthful for mouthful, we’re consuming roughly the same amount of sugar whether we’re enjoying a drink or a pastry.

Sugar per mouthful of coffee drinks and pastries from Starbucks. Assuming that a mouthful of liquid is ~31.6 grams and a mouthful of baked good is ~7 grams. Mouthful sizes are approximate and based on my own “field research.”

If these drinks and pastries really do have comparable amounts of sugar in each bite, why do I consistently find my first swig of a chai latte to be sweeter than my first bite of a blueberry muffin?

Perceived sweetness is more than just sugar

I think that much of the answer is that sugar consumption does not equal sweetness perception.

The scientific literature suggests that humans perceive sweetness differently in solids and liquids: in general, experimental participants thought that liquids were sweeter than solid foods3. In addition, the sugar’s distribution throughout the food may matter. One study suggests that a bite with chunks of sweetness seems sweeter than a bite with a homogenous level of sweetness, even when controlling for the overall sugar content of the bites4. Interestingly, this effect could make baked goods taste sweeter than drinks – my bites of a chocolate chip cookie have a fairly heterogenous sugar content, whereas my swallows of a chai latte should be pretty homogenous.

The presence of other flavors also matters. If you’ve ever tried baking brownies without the chocolate, you’ll likely find that the result seems sweeter than the chocolatey version, even though the overall sugar content has likely fallen5. And the literature supports this idea – a small amount of salt seems to heighten sweetness, while sour and bitter flavors dampen it6.

If we take this investigation out of the lab and into the wild, do baked goods have more of these balancing flavors than coffee shop drinks? This is a harder question to answer definitively, because we don’t know Starbucks’ actual recipes. But my intuition is that baked goods do generally contain more balancing flavors. A typical baked good contains salt, fat, fruit, chocolate, spices, and other ingredients to balance out the sugar. Drinks also have some of these – the flavors from tea, coffee, dairy products, spices, and chocolate could all be balancing out a sugary drink as well. Perhaps the key is whether these flavors are used enough to effectively balance out the sugar. In my experience, something like a cardamom latte tends to taste mainly of sugar, with only a faint whisper of cardamom.

Why is everything so sweet, anyway?

In the end, my biggest takeaway from Starbucks’ nutrition information wasn’t that their drinks are sweeter than their baked goods. It was that everything they sell is extremely sweet. Unhealthily sweet. According to current health recommendations, adults shouldn’t have more than about 30 grams of sugar per day7. So, drinking 8oz of a sweetened Starbuck’s drink (17 grams of sugar) provides half of your sugar allotment, eating a baked good (24 g. sugar) gets you close to your total allotment, and sitting down to a drink and a baked good (41 grams of sugar) exceeds it.

I opened this piece by observing that coffee shop pastries aren’t too sweet, but their drinks are. But upon further reflection, I have to walk that claim back. I was indexing too much on my local bakeries and coffee shops, whose baked goods do seem less sweet than Starbucks’. To confirm this suspicion, I ran a totally-biased test: I tasted a blueberry muffin from Starbucks side-by-side with one from my favorite local bakery.

The Starbucks muffin indeed seemed sweeter. Specifically, it seemed more monochromatically sweet, without other flavors to offset that sugariness. It did have a lemony flavor, but mostly it just tasted sweet. In contrast, my local bakery’s muffin was nuttier and tasted more like blueberries, butter, and a hint of cinnamon.

So, mass-market coffee shops like Starbucks sell incredibly sweet products, both solid and liquid. Why? Wouldn’t their products be tastier, and sell better, if they used less sugar and more flavor?

I suspect that part of the reason is that the American palate is just very acclimated to sweetness. Sweetened drinks are ubiquitous. Pumpkin Spice Lattes even have a season, like truffles or hunted game. And of course, there’s soda. Even diet sodas with no actual sugar still taste sweet. Embedded within this sweet landscape, perhaps a caramel brulée latte with 24 g. of sugar simply doesn’t seem terribly sweet. One (under-powered) 2015 study supports this notion that our palates adapt to our habitual sugar intake: foods tasted sweeter (or too sweet) to people who had cut out all sweeteners for 2 weeks8.

But interestingly, the coffee drinks at my local bakeries do still taste too sweet to me, even if their pastries mostly don’t. I would expect these to be at a similar level of sweetness – if the same population of customers is ordering chai lattes and blueberry muffins, shouldn’t the bakery try to calibrate both pastries and drinks to a similar level of perceived sweetness? Perhaps the bakery is actually serving two distinct populations: one that orders sweet drinks, and the other that orders sweet pastries. Or perhaps the drinks remain super-sweet because syrup is just a cheap, shelf-stable way to deliver flavor to a drink – you could make a hazelnut latte by infusing the milk with ground hazelnuts, but using pre-made hazelnut syrup is much quicker, doesn’t require extra preparation, and won’t spoil. Because of that it’s probably also cheaper.


  1. Table 1

    ↩︎
  2. The Centers for Disease Control recommends consuming <12 teaspoons, or 50 g. of sugar per day; the American Heart Association says 36 g. for men and 25 for women; the UK’s National Health Service says <30 g.↩︎

  3. Shimada et al., 1990; Calviño et al., 1993; Alley & Alley, 1997↩︎

  4. Kistler et al., 2021↩︎

  5. If your brownie recipe uses unsweetened cocoa powder, this experiment won’t change the overall sugar content but will increase the sugar’s density. If it uses solid chocolate (which contains some sugar), it will actually reduce the sugar.↩︎

  6. Shimada et al., 1990↩︎

  7. see note 2↩︎

  8. Bartolotto (2015)↩︎

Avatar
Leyla Tarhan

PhD in Cognitive Neuroscience, making a move into industry.

Related