Trace my four-part journey through more than 20 select batches of vegan macarons changing one variable at a time to develop the best recipe.

Part I

In a recent recipe review, I scrupulously followed three vegan macaron recipes:

• Tom Adams, AKA Project Vegan Baking
• Chloe Miller-Kelly, AKA Baked By Clo
• Camila Hurst, AKA Pies and Tacos

By far, after dropping my oven 10°F (5-6°C), Hurst’s recipe was most successful for me. I thus made it the base recipe for developing my own.

Because while it was most successful, some macarons were lopsided and all were hollow – even after 48 hours of maturing. I reasoned my airflow and meringue needed help.

I thus introduced three variables, separately and together, into Hurst’s recipe:

• reduced aquafaba
• 1:1 aquafaba to granulated sugar ratio
• flipped tray

Theoretically, the first two variables would improve meringue structure; third, airflow. I was certain I’d have my own recipe after testing…

The Method

I try to run all related tests in a day. Conducting several, hours-long macaron trials, however, is not possible for my business of one. Nevertheless, I controlled what I could:

• Environment – I cleared an area near a dehumidifier where macarons could rest between 45-55% relative humidity.
• Equipment – All tools were meticulously scrubbed with a new sponge, dried and rubbed with vinegar using a freshly cleaned lint-free towel, and air-dried before and between tests. Per Hurst’s recommendations, I used two Nordicware aluminum half sheet pans and two Aprince silicone baking mats. (These are not affiliate links.)
• Ingredients – Every test batch required aquafaba, cream of tartar, granulated sugar, almond flour, and confectioner’s sugar. I sourced the following US-nationally available products:
  • Eden Organic Garbanzo Beans
  • Great Value Cream of Tartar
  • Zulka Morena Pure Cane Sugar
  • Bob’s Red Mill Almond Flour
  • Baker’s Corner Powdered Sugar

The Madness

What do you do with failed batches?

After years of wrestling with this idea, I settled on throwing them out.

I save 2-4 shells to check texture after maturation, but for the most part, failed batches go in the trash can.

I used to do everything in my power to repurpose even the soggiest, crispiest, or otherwise subpar macarons. Yet after a while, I realized I was using more food and time trying to elevate something I genuinely didn’t want – before or after transformation. (Soggy macaron marzipan and crispy macaron chocolate bark aren’t that good.)

Logically, I began turning toward the failed macarons directly, without doing anything to change them. They were perfect for eating my frustration! Until I realized:

I’m using my body as the trash can.

So now, into the actual trash can they go.

You’re still wasting money!

You’re not wrong, but I have a theory:

A failed batch of macarons is more temporally and emotionally expensive than it is financially so.

At the time of this writing, the in-store expense for each ingredient was as follows:

PRODUCT	PRICE	NET WEIGHT	AMOUNT in RECIPE	PRICE of AMOUNT in RECIPE
Eden Organic Garbanzo Beans*	$4.89	29 oz (822 g)	75 g or 150 g	$0.45 or $0.89
Great Value Cream of Tartar	$2.47	2.75 oz (78 g, or 6 Tbsp/90 mL)	¼ tsp (1.25 mL)	$0.03
Zulka Morena Pure Cane Sugar	$6.95	5 lb (2.72 kg)	66 or 75 g	$0.17 or $0.19
Bob’s Red Mill Almond Flour	$10.99	1 lb (454 g)	110 g	$2.66
Baker’s Corner Powdered Sugar	$2.09	2 lb (907 g)	110 g	$0.25

*One could argue there’s no expense for aquafaba as it’s typically destined for the drain.

Therefore, each batch cost me $3.56-$4.02 if counting aquafaba; $3.11-$3.13 if not. This clearly isn’t nothing! But I’m grateful to be in a place where I can say around $3.50 a batch is just about worth what I can learn from each.

As for equipment, I had everything already. Macarons don’t require specific tools, and since I’ve been an avid baker for years, even piping tips (the most specialized item they require) were in my collection.

Is Baker’s Corner Powdered Sugar vegan?

Truthfully, I’m uncertain. In conjunction with my family and PETA, I’ve made an informed decision to prioritize our budget, and in turn, my ability to bake often.

Organic sugar is undoubtedly processed without bone char, but unfortunately more expensive. If I ever see a day when money is no object, I may reassess my grocery habits. For now, I find comfort in this quote:

“Don’t stress too hard about sugar if you’re unsure about how it was produced.”

PETA continues to say the average vegan already saves around 200 animals per year. Besides, they suggest the best way to attract others toward this lifestyle is to stress a little less sometimes. Sugar, for me, is one of those times.

The Tests

In addition to the aforementioned constants, my dry mix of 110 g almond flour and 110 g powdered sugar remained the same across trials. I also always used 75 g aquafaba, reduced or otherwise.

ROUND 1: Single Variable

Batch 1

unreduced aquafaba	control
66 g granulated sugar	control
flipped trays	variable

Summary: bad, but evenly bad

Hypothesis

Flipped trays indeed help with airflow. While the feet of these shells spread, they only spread slightly and evenly. Each looked like a miniature target with a domed top as the bull’s eye and a spread foot as the next best concentric circle to hit.

Batch 2

unreduced aquafaba	control
75 g granulated sugar	variable
regular-facing trays	control

Summary: mostly pretty, but fragile

Hypothesis

The aquafaba:granulated sugar ratio is en pointe at 75:66 (or 15:22 for my oldest who likes to remind me to reduce). These shells were pretty to behold! Yet they were so fragile, I lost several to the routine acts of lifting them off my trays and sandwiching them with “buttercream”.

Batch 3

reduced aquafaba	variable
66 g granulated sugar	control
regular-facing trays	control

Summary: fuller, but lopsided

Hypothesis

A stronger meringue is needed to uphold the 220-g dry mix. Many macaron recipes – both vegan and conventional – are at or around 1:1:1:1 for aquafaba (or egg whites) to granulated sugar to almond flour to powdered sugar. Hurst’s is dry-heavy at a rounded 8:7:12:12.

ROUND 2: Double Variable (Single Control)

Batch 4

unreduced aquafaba	control
75 g granulated sugar	variable
flipped trays	variable

Guiding Question: Do flipped trays cure fragility?

Short Answer: No

I honestly had little reason to think flipping trays would cure hollows. I tried it because a small part of me believed that improved heat distribution would improve ingredient distribution too. It does not; Batch 4 was as fragile as Batch 2.

Batch 5

reduced aquafaba	variable
66 g granulated sugar	control
flipped trays	variable

Guiding Question: Do flipped trays cure lopsidedness?

Short Answer: Yes

It does! Twice now, after baking on flipped trays, I had a mere 1-2 lopsided shells (not counting the one I accidentally smashed with the thumb of my oven mitt).

Batch 6

reduced aquafaba	variable
75 g granulated sugar	variable
regular-facing trays	control

Guiding Question: Do flipped trays help structure at all?

Short Answer: No

No matter if I flipped my trays or reduced my aquafaba, increasing granulated sugar increases fragility. Batches 2, 4, and 6 were equally difficult to remove from my mats without breaking, and even trickier to sandwich.

ROUND 3: All Variables (No Controls)

Batch 7

reduced aquafaba	variable
75 g granulated sugar	variable
flipped trays	variable

Cancelled: not worth proceeding

Hypothesis

If flipped trays have no effect on structure, and 75 g granulated sugar produces fragile macarons no matter what, then combining these variables is a waste of resources.

Conclusion

Reducing aquafaba and baking on flipped trays are worthwhile; adding more granulated sugar is not. Though reduced aquafaba has little effect on hollows (and flipped trays only address lopsidedness), an 11-g uptick in granulated sugar is detrimental overall.

My base recipe in the next few rounds thus involves 75 g reduced aquafaba, 66 g granulated sugar, flipped trays…and more pictures.

Part II

In Part I, Batch 3, reduced aquafaba produced slightly fuller shells. I logically concluded an even stronger meringue would lead to even fuller shells.

Yet other sources dared to suggest adding more almond flour to my already dry-leaning mix! I therefore chose three new variables for this round: two that would tighten my meringue, and one that would tip the wet:dry ratio harder towards dry.

The Variables

Mock Swiss meringue

In conventional cooking, a Swiss meringue means mixing egg whites and granulated sugar over a double boiler until the mixture reaches 160°F/71°C (at which point the sugar is also dissolved). The goal is not only to pasteurize the egg whites, but also to produce a tighter meringue by eliminating those aerating sugar granules.

Mine is a mock Swiss meringue because I disregard temperature (not to mention it’s vegan). Chickpeas needn’t ever be pasteurized! Immediately after reducing, I add granulated sugar to my hot aquafaba and stir to dissolve.

½ tsp (2.5 mL) cream of tartar

In its own way, cream of tartar also produces stabler meringues. I was hopeful my recipe simply needed an acidic boost. (Spoiler Alert: Cream of tartar alone solves little.)

125 g almond flour

A heftier dry mix was my wild card variable! I was not convinced this was my problem at all, but too many people suggested I try it for me not to.

More Madness

On a less consequential level, I added a fourth variable: a different brand of almond flour. I buy Bob’s Red Mill almond flour at this odd discount store near me that carries random overstock and, somehow, a consistent supply of Bob’s Red Mill.

But even at this discount store, BRM almond flour was costing me $10.99 per 1-lb (454-g) bag – and I was running out of coupons. So I switched to Baker’s Corner (Aldi brand) almond flour where the same size bag was $5.29 in-store.

Surprisingly, it’s better quality! It’s drier and lighter, and because it’s significantly cheaper, it brought my batch price down from $3.56-$4.02 to $2.37-$2.58 (including aquafaba; $1.48-$1.69 if not).

The Tests

Previously stated controls around environment, equipment, and ingredients were repeated here. Moreover, each of the following batches contained 75 g reduced aquafaba, 66 g granulated sugar, and 110 g powdered sugar. All were baked on flipped trays.

ROUND 4: Single Variable

Batch 8

½ tsp (2.5 mL) cream of tartar	variable
French meringue	control
110 g almond flour	control

Batch 9

¼ tsp (1.25 mL) cream of tartar	control
mock Swiss meringue	variable
110 g almond flour	control

Batch 10

¼ tsp (1.25 mL) cream of tartar	control
French meringue	control
125 g almond flour	variable

ROUND 5: Double Variable (Single Control)

Batch 11

½ tsp (2.5 mL) cream of tartar	variable
mock Swiss meringue	variable
110 g almond flour	control

Stabilizing mock Swiss meringue with ½ tsp (2.5 mL) cream of tartar was not the solution I hoped it would be. Though my meringue needed help – and the mock Swiss method helped more than ½ tsp (2.5 mL) cream of tartar – it did not need this double punch.

The feet of this batch spread, and the lacey bottoms were gummy from too much lift and not enough structure. That said, there was structure! Some of that lace was connected to the tops of my shells. This batch made me think, “Perhaps more almond flour is needed after all…”

Batch 12

¼ tsp (1.25 mL) cream of tartar	control
mock Swiss meringue	variable
125 g almond flour	variable

These shells were among my fullest yet (beaten only by Batch 10). However, they were quite tall and slightly spread. I also noticed, for the second time, 125 g almond flour leads to grainy tops. 

Besides, all that dry mass twice made the macaronage significantly harder to read. I seriously struggled to know when to stop mixing this batch as well as Batch 10.

I planned to decrease almond flour to 115-120 g, and debated increasing powdered sugar to 115 g as well. Perhaps the former would reduce graininess and excessive difficulty during macaronage without sacrificing fullness, while the latter would smooth tops.

Intermission

At this point, Batches 8-12 were mature. I cut them in half and ranked them from fullest to most hollow:

1. Batch 10: Single Variable, 125 g almond flour*
2. Batch 12: Double Variable, mock Swiss meringue + 125 g almond flour*
3. Batch 9: Single Variable, mock Swiss meringue
4. Batch 11: Double Variable, ½ tsp (2.5 mL) cream of tartar + mock Swiss meringue*
5. Batch 8: Single Variable, ½ tsp (2.5 mL) cream of tartar

I immediately noticed the best batches contained 125 g almond flour; worst, ½ tsp (2.5 mL) cream of tartar. Does the extra cream of tartar need the extra almond flour (or vice versa)?

I concluded it does not. The batches with asterisks indicate batches that slightly spread. I think in all cases, whether due to extra almond flour or a doubly tightened meringue, I was forced to overmix in order to achieve the proper piping consistency. Those batters were thick!

I thus chose to conduct my next two tests with ¼ tsp (1.25 mL) cream of tartar and, after an embarrassingly long internal debate sustained by too much Google, 115 g each almond flour and powdered sugar. The only variable, therefore, was the meringue style: one French, one mock Swiss.

Round 6: Battle of the Meringues

In addition to 66 g granulated sugar, the final two batches of Part II both included ¼ tsp (1.25 mL) cream of tartar and 115 g each almond flour and powdered sugar.

Batch 13 (French Meringue)

Observations: A 10- rather than 15-g increase in dry mass still made for a trickier macaronage. Nevertheless, I don’t think I overmixed as I stopped just as my batter loosened into something about as viscous as my hair conditioner. I may have even under-mixed!

Yet my feet were ruffled and spread on both trays. I suspected under-resting after the first tray was done, but then the second tray (which had rested for an additional half hour while the first baked) looked the same.

The last technical culprit could have been high oven temperature. However, I was solidly at 275°F (135°C) which has proven to be my sweet spot.

I’m therefore not entirely sure what went wrong. Perhaps 120 g almond flour would have been better? (My best batch to this point was French meringue with 125 g.)

Batch 14 (mock Swiss meringue)

Observations: I had 6 absolutely perfect shells! And, perplexingly, 42 lopsided and spread ones.

Like my French batch, I’m not entirely sure what went wrong. My meringue was stiff; macaronage, slightly under (the least egregious mistake one can make with vegan macarons); resting, just right; oven, stable at the usual temperature.

I suspected these tighter batters have been making the macaronage difficult to track; more dry mix means more in the way of seeing how much the meringue deflates with each fold. Maybe I was way under- (or over-) mixed after all?

Yet more than that, I wondered if my oven temperature was indeed too high for the size shells I’d been piping. Each of my mats has two concentric circles for 24 macarons. I’d been piping to the smaller circle, but if I piped closer to the larger one, would I have more batter per shell to withstand 275°F (135°C)?

Conclusion

I concluded that I had no conclusion at this point. French meringue looked more consistent, but mock Swiss looked more promising.

I needed a tie-breaker (or two).

Part III

After wrestling with how to adjust my dry mix again, I realized I never tested decreasing granulated sugar! Increasing from 66 to 75 g made for impractically fragile shells in Part I. Would 55 g do the opposite?

The next two batches thus test 55 g with French and mock Swiss meringue. For a fair comparison, I returned to 110 g each almond flour and powdered sugar.

Monkey Wrench

I ran an errand to buy more Eden Organic Garbanzo Beans. For the best price per unit during testing, I’d been buying 29-oz (822-g) cans.

Dismayingly, a sign greeted me with an apology for delayed shipments and low stock. There were no 29-oz (822-g) cans, and only 4 15-oz (425-g) ones. I needed 6.

So I bought those 4 plus 2 tetrapacks of 365 Organic Unsalted Garbanzo Beans. I knew adding another variable was premature, but I felt close enough to victory to risk yet another brand swap. Plus, at $1.39 per 13.5-oz (380-g) box, this aquafaba cost me $0.55 instead of $0.89 per batch.

The Tests

Round 7: 55 g Granulated Sugar

Batch 15 (French Meringue)

By far, these were the fullest macarons I’ve made with French meringue! Unfortunately, most of them were lopsided, and all of them looked chalky.

I expected less sugar to mean less sheen. But these were so matte, they looked like I overbaked them – even though I hadn’t.

Maybe I didn’t rest them long enough? The tight, lower-sugar meringue with its lack of shine made this stage difficult to judge both by feel and look. Moreover, I’m not entirely confident in my macaronage; the goldilocks stage of mixing is more elusive with these stiffer batters!

Yet lopsidedness is a symptom of over-resting. Perhaps my meager 30-minute rest was actually too long? Batch 14 (mock Swiss with an increased dry mix, 115 g each almond flour and powdered sugar) was also mostly lopsided. I wonder if these tighter, drier batters can go directly into my oven…

Batch 16 (Mock Swiss Meringue)

Based on the adjacent French meringue test, I decided to take a wild risk: no rest.

If this was destined to work, my macaronage had to be absolutely perfect. I still wasn’t sure what that looks like with a stiffer batter, so I filled one piping bag after 70 folds and the other after 85.

I was shocked! The 85x-folded batter worked beautifully producing 11 of 13 sellable shells – even, smooth, full – versus only 1 of 14 with the earlier half.

More folds and less rest seem to be key. Less granulated sugar, however, has a clear downside of unpleasantly chalky, gritty tops. I think 66 g granulated sugar to 75 g reduced aquafaba is too high, but 55 g is too low.

I wonder: How does each meringue respond to 60 g granulated sugar and no rest?

Round 8: 60 g Granulated Sugar, No Rest

Batch 17 (French Meringue)

I did it! I made my first batch of completely sellable shells. They puffed quite a bit, but the feet are gorgeous. I would proudly sell these.

The granulated sugar content is just right too. I didn’t entirely mind the puff because 60 g and no rest allowed for gently lustrous, smooth domes that deserved the extra space they claimed.

The only downsides are that they used 120 g almond flour, the most expensive ingredient, and took 38 minutes to bake. Mock Swiss uses 115 and has consistently needed less time. I’m hopeful the next trial will be just as good as or better than this one; if anyone plans to mass produce these shells, those few grams and minutes can really add up!

Batch 18 (Mock Swiss Meringue)

And it is! Perhaps due to a tighter meringue, these shells didn’t balloon as much as the French batch making them look more like genuine Parisian macarons. They are beautiful and full.

Conclusion

My concerns now are around food coloring. How does this recipe respond to being a culinary sandbox?

Part IV

I can only say I’ve found my recipe if it can withstand some creative energy.

The following tests are straightforward. Foremost, I needed to know if I could repeat my success. My plans thereafter were to add food coloring, and if successful, test that again too.

The Tests

Round 9: Mock Swiss Playground

Batch 19-21 (Use Different Aquafaba; Add Food Coloring.)

At this point, I was forced to switch to 365 Organic Unsalted Garbanzo Beans. It reduced and chilled enough like Eden Organic that I decided it was safe to proceed with an additional variable: food coloring.

And it was safe, successful even, though this aquafaba is definitely thinner. Every shell looked fantastic! However, I needed to bake them for 35 minutes for the bottoms to peel off my mat as cleanly as they did in Batch 18.

Batch 22 (Swap 7 g Powdered Sugar for Cocoa.)

I have found my recipe! Making small swaps and still having success is one of the major milestones I wanted to hit.

I still stand by Eden Organic. 365 yields a slightly weaker meringue, and in turn, slightly weaker macarons. I believe the difference is caused by the presence of kombu seaweed in Eden Organic. This ingredient supplies sodium alginate, a gelling agent, which bolsters the ensuing meringue’s ability to hold its shape.

I thought about reducing 365 aquafaba with a piece of kombu seaweed, but this thought opened multiple doors of questions and tests that left this sole proprietor feeling overwhelmed. How big of a piece? Should I offset any liquid the dried seaweed will absorb? How much will it absorb? Should I soak it first? Will it lose too much of its gelling power through soaking? Will it work at all?

I’ll stick with Eden Organic 15-oz (425-g) cans for now. The store-imposed swap I described in Part III led me to discover that the aquafaba in their 29-oz (822-g) cans is diluted, barely better than 365.

Conclusion

Stay awhile to peruse my intricate vegan macaron recipe distilling the best information herein alongside other discoveries not explicitly mentioned (like the time I tried – and failed – to whip previously homemade and frozen aquafaba).

There are still questions to be answered, but for now, I have hit my goal of developing a reliable formula. Future explorations include:

• How does my recipe respond to extracts or other flavoring agents?
• Is this batter stiff enough for special shape and character macarons?
• Would powdered aquafaba be just as effective, and perhaps more reliable, than fresh aquafaba from Eden Organic cans? If so, how should I rehydrate it?

May my discoveries to this point empower you to troubleshoot your own macaron journey. Reducing your aquafaba, lowering your granulated sugar, increasing your dry mix, flipping your trays, and decreasing your rest time and temperature may also be the answers you need – regardless of whose recipe you’re following.

Because in the end, baking macarons is a highly personal endeavor of precision cooking in your specific kitchen. No one else’s experiments can fully explain your unique struggles with these cookies; I just hope that mine can help.