The Theory and Science Behind Coffee Roasting
Coffee roasting sits at the intersection of agriculture, chemistry, and sensory evaluation. Long before water meets ground coffee, roasting has already defined the majority of what ends up in your cup; acidity structure, sugar development, aromatic compounds, body. But here's the thing that's easy to overlook: everything that makes a coffee great is already in the seed when it arrives at our roastery. We're not adding flavor. We're revealing what's already there.
As roasters, we’re handling coffee near the very end of the process. Where a producer may be working the better part of a year on a single harvest (managing soil and shade, plant health and disease, hoping the weather cooperates, harvesting, processing, drying and preparing for the following harvest, etc.), a roaster spends a relatively small amount of time with the coffee (everything we’ll go over below). Although roasting can be complex, the risk level and time commitment is comparatively much lower.
What Roasting Actually Is
Roasting is the controlled application of heat to transform dense, green coffee seeds into an aromatic, soluble bean. Sounds straightforward, but that one sentence hides a lot of complexity. Heat transfer mechanics, bean chemistry, moisture content, water activity, processing method, density, equipment design. All of it influences how a coffee behaves in the roaster and how it tastes in the cup.
Green coffee is chemically and physically different from roasted coffee in almost every way. Inside the seed you've got complex carbohydrates, simple sugars, amino acids, organic acids, lipids, and water; typically around 9–12% by weight.
Once heat is applied, a cascade of overlapping reactions occur. Maillard reactions (the same reactions seen when searing steak or toasting bread) happen when sugars and amino acids collide, forming hundreds of aromatic compounds and browning pigments. Caramelization thermally degrades sugars, contributing sweetness and body. Strecker degradation converts amino acids into aldehydes and other volatile aromatics. And all the while, internal steam pressure is building until it drives First Crack, a structural fracture event where the cellular walls physically rupture. Basically the same mechanism as popping popcorn.
Past First Crack, continued development increases solubility and roast character but starts eroding origin, varietal, and processing clarity. That trade-off is at the center of every roasting decision we make. The goal for single origins, is to develop just to the point of enhanced sweetness while highlighting origin characteristics and avoiding roast character. Our blends are developed a bit more than our single origins as we’re emphasizing sweetness, body and comfort in these coffees.
So roasting isn't just making coffee darker. It's managing heat and time to control how and when all of these reactions happen.
A Closer Look at Each Phase
Phase One: Drying
Green coffee (at 9–12% moisture by weight) needs to be dried evenly before anything else can happen. As the seed heats up, internal moisture converts to steam and starts escaping through the cellular structure. Beans go from green to pale yellow, and the aroma shifts from fresh-cut grass to warm grain.
This phase matters more than people tend to think. Push too hard early and you get tipping (burning of the embryo) and the outside of the bean races ahead of the inside. Do this and you end up with coffee that tastes hollow and somehow and both over-roasted and underdeveloped; ashy, but with a sharp acidity and dry astringency. Go too slow and the whole roast drags, and you get something lacking sweetness, vibrancy and structure. A well-managed drying phase sets the trajectory for everything after it.
Phase Two: Maillard Reaction
Once the coffee moves past yellowing, it enters the Maillard phase (browning reactions between sugars and amino acids). This is where things start to get interesting. Aroma picks up fast: baked bread, toasted nuts, light caramel. Bean color deepens from yellow to light brown as hundreds of flavor compounds form. Sugars are breaking down and recombining. Acidity is evolving and integrating. Body and texture are taking shape.
How fast or slow you move through this phase has a direct effect on balance. A longer Maillard phase tends to emphasize sweetness. A shorter one tends to push floral and fresh fruit notes forward. This is where the coffee's complexity and structure really develop.
Phase Three: Development
Development starts at First Crack, that audible pop when the bean's cellular walls fracture under steam pressure. At this point the bean becomes porous enough to extract properly when you brew it.
From here, sweetness deepens, acidity softens, and roast tones (cocoa, browned sugar, spices), if you go far enough, start to show up. For our lighter-roasted coffees (all single origins, some blends), we keep development intentionally short: typically 8–10% of total roast time. We want to spotlight varietal character, processing nuance, and terroir. The longer you let the roast go and the higher the end temperature, the more roast character takes over as the dominant flavor.
Three Roasting Contexts in Our Workflow
In our roastery, we run two primary roasting systems: the Roest L200 Plus sample roaster and the Loring S35 Kestrel production roaster. They play very different roles in our process, and we'll get into both.
1. Sample Roasting — Meeting a Coffee
Sample roasting is how we get our first real read on a coffee.
When offer samples come in from producing partners, we start analysis right away. This step matters a lot. A botched sample roast can mean we pass on something exceptional because our mistake kept it from showing what it could do and a blown green analysis can mean we purchase something we shouldn’t.
Before we roast, we evaluate green quality, density, and moisture. After roasting, we measure weight loss and whole bean and ground color, then dig into the roast data to understand how the coffee transformed. The important thing here is that sample roasting is intentionally neutral. We're going for clarity, not style.
Then the sample goes to the cupping table, where it's evaluated blind alongside other offers. We're assessing quality, looking for cup defects, mapping flavor profile, and assigning a score with tasting notes. Physical and sensory analysis together are a major factor in whether we buy a coffee, though not the only one.
2. Profile Design Roasting — Learning the Coffee
Once we've purchased a coffee and it arrives at our roastery, we move into profile development, still on the Roest. No two coffees roast the same. Physical and chemical differences require a distinct approach every time.
This is where roasting gets investigative. We're testing and recording charge temperature, rate of rise (measurement of thermal gain) trajectories, turning point timing, drying phase duration, Maillard duration, development time ratio, end temperature, weight loss, and color. The Roest lets us run through a lot of different approaches quickly and with minimal waste, so we can build a real picture of how a coffee responds to heat.
Some coffees reward aggressive early energy. Others fall apart under it. Some need a longer Maillard phase to unlock their sweetness. Others lose vibrancy if you hold them there too long. Profile design isn't so much about imposing a style, it's more about reading what the coffee's physical and chemical signals are telling you and responding to that.
Most coffees take 3–4 profile design roasts (varying heat application, phase modulation, and end temperature, color and weight loss) before we have a clear direction. Then it's back to the cupping table. Some coffees take several rounds before we get where we want to be. We try to start this process at least 2 weeks before releasing a new coffee.
3. Production Roasting — Taking It to Scale
Once we know how we want to treat a coffee, we have to figure out how to do that on the Loring, which handles 60–70 pounds per batch versus 100–185g on the Roest. The engineering and heat transfer properties of the two machines are fundamentally different.
The specific numbers (temperatures, gas settings, times, weight loss) won't transfer directly. But the phase structure, color targets and weight loss we're aiming for, all of that carries over. We build a roasting theory for the Loring based on what we learned on the Roest, run the first production batch, cup it, and adjust from there.
After that first production roast, we record weight loss and color readings, analyze the roast data, and cup the coffee. Then we decide whether to proceed or change something.
A note on numbers:
We intentionally avoid citing specific roast times or temperatures. Not because we're guarding secrets but because those numbers are machine-specific. What happens at a given time and temperature on our equipment won't reflect what happens on yours.
If you're learning to roast, resist the urge to match someone else's profile by time and temp. It’ll lead you down a long and unproductive rabbit hole (as young roasters, almost 20 years ago, we didn’t resist the urge and after A LOT of frustration figured out this isn’t the way). Focus instead on percentage of time spent in each phase, weight loss, roast color and end temperature relative to your own system and your own past roasts, not compared to someone else's roasts.
The same goes for color readings. Even identical make-and-model color analyzer/grinder setups produce different numbers. We've sent coffee to industry friends running the same analyzer/grinder setup we use and recorded readings as much as 5 points apart on the same coffee. That's a massive difference. So don't try to hit your favorite roaster's color numbers. Use color as a data point relative to your own roasts and what you're tasting.
Because the Loring and Roest both transfer energy through convection, we have found that weight loss and color are consistent data points in our system. Meaning a coffee roasted to a ground bean color of 105 with a weight loss of 10.5% on the Roest will taste very similar to a coffee roasted to the same specs on the Loring. But, matching weight loss between machines can lead you down the wrong path. If we were to match these numbers between a coffee roasted on a direct flame drum roaster, such as a vintage Probat, and a Loring, the cup profiles would differ. Different equipment, different thermodynamics, different approaches, different results.
The most useful thing you can do is collect your own data, get good at tasting, and build an understanding of how modulating a roast, and the resulting changes in roast phases, weight loss and color, affects what's in the cup.
Why We Roast Light
Our roasting philosophy comes down to transparency to origin. We want to amplify what makes each coffee distinct while enhancing its natural sweetness. Clear, vibrant, defined and sweet is the target.
When we say "light roast," we don't mean underdeveloped. We mean fully developed, just without unnecessary roast character layered on top.
Lighter roasting preserves the things that made us want to buy the coffee in the first place: organic acid structure, floral aromatics, varietal expression, processing character, terroir. Darker roasting increases solubility and body, but it trades all of that for roast-derived flavor: smoke, carbon, bittersweet compounds. For commodity coffee, that trade-off can mask defects and create consistency. For specialty coffee, it just covers up what makes each lot special.
We put real effort into sourcing differentiated coffees from producers doing specific, intentional work. Roasting in a way that erases those differences would defeat the purpose.
Closing
Roasting is applied thermodynamics guided by sensory feedback.
Data informs every decision: rate of rise, development time, charge temperature, end temperature, weight loss, color. But the data doesn't tell you if the coffee is good. The cup tells you that.
The two-machine system, the rounds of profiling and cupping, the attention to phase timing, it all serves one objective: getting each coffee to the point where origin, processing, and natural sweetness come together as clearly as possible. That's when coffee tastes most like itself.