Save up to 50% in our Annual Sale
Countdown has ended
Roasting is not 'light', 'medium', or 'dark'. It is a controlled application of heat over time, tracked by data, guided by sensory feedback.
A roast profile is not a colour target. It is a record of every energy decision made during a roast, and the bean's response to those decisions.
Understanding roast curves does not require access to a roasting machine. It requires understanding what the data represents: how heat is applied, how the bean responds, and how each phase of the roast affects the chemistry and flavour of the finished coffee.
This is the difference between roasting as a craft informed by data and roasting as a reactive process guided by colour and instinct.
What a Roast Curve Shows
A roast curve is a graph of bean temperature (or environment temperature) over time, typically displayed on roasting software in real time and stored for analysis and replication.
The key features of a standard roast curve include charge temperature, the drum temperature when green coffee is loaded, which varies by roaster design, batch size, and bean density (too high risks scorching, too low means excessively slow early heating); the turning point, the temperature minimum on the bean probe as the cold green coffee absorbs enough energy from the drum environment to stop cooling the probe; Rate of Rise, the speed of bean temperature increase, one of the most watched metrics in modern specialty roasting; the yellow point, when beans visually transition from green to yellow and the Maillard reaction begins to accelerate, typically at 140–160°C bean temperature; first crack, marking the transition to development phase; and end of roast or drop temperature, the point at which the coffee is discharged to the cooling tray.
Rate of Rise: The Critical Metric
Rate of Rise (RoR) describes how quickly bean temperature is increasing at any given moment. It is typically a declining curve through most of the roast, starting higher in the drying phase and progressively declining as the roast proceeds.
Flick (RoR Increasing)
If RoR increases suddenly, particularly late in the roast, it indicates a sudden input of energy that was not anticipated. This 'flick' can cause uneven development: the exterior of the bean accelerates while the interior cannot keep up. It is a significant roast flaw that shows in the cup as harsh or aggressive character.
Crash (RoR Dropping to Zero or Below)
If RoR crashes, temperature rises stop or reverse, the roast has stalled. Insufficient energy is being applied for the reaction chemistry to continue at the appropriate rate. The result is 'baking', a roast defect discussed in the next post.
Skilled roasters aim for a steady, declining RoR, continuously slowing, but never stalling. This requires continuous energy adjustments, gas, airflow, drum speed, throughout the roast. Modern roasting software makes this visible in real time; the discipline is interpreting and responding to the data accurately.
Development Time and DTR
Development Time (DT) is the duration from first crack to the end of the roast. Development Time Ratio (DTR) is development time expressed as a percentage of total roast time. These metrics reflect how much of the roast was spent in the post-first-crack phase, the period of most active structural transformation and flavour development.
Short Development Time
Preserves more acidity and aromatic volatiles, highlights origin character more directly, produces lighter colour and lower solubility, and risks underdevelopment if too short.
Longer Development Time
Reduces perceived acidity as chlorogenic acids degrade further, builds sweetness from continued sugar caramelisation, increases body and mouthfeel, and shifts the flavour balance toward roast-driven character over origin character.
DTR in specialty roasting typically ranges from approximately 20–25% for lighter profiles to 25–30% for medium profiles. Very dark profiles extend beyond this range.
DTR is a useful metric but not a prescription. The correct DTR depends on the bean density, moisture content, batch size, roaster design, and target cup profile. A DTR that produces excellent results with a high-density Ethiopian natural at 2,000m may produce underdeveloped or baked results with a lower-density Brazilian natural at 900m.
Total Roast Time
Total roast time, from charge to drop, affects the overall heat exposure of the coffee and the balance between different reaction stages.
Shorter total roast times, 7–10 minutes in drum roasters, shorter in fluid bed, produce what some roasters describe as more 'transparent' cups, where origin character is foregrounded. Longer roast times, 12–16+ minutes, allow more gradual heat transfer and, in some cases, more integrated flavour development, though extended roast times at insufficient heat create baking defects.
There is no universally optimal roast time. The interaction between total roast time, development time, and RoR profile is what matters, and these are assessed in combination, not in isolation.
Replication and Consistency
One of the primary functions of roast curve software is roast replication.
A single exceptional roast of a new lot tells a roaster how that coffee can taste at its best. A saved, replicable profile tells them how to produce that result consistently across every subsequent batch of the same lot.
Replication requires controlling variables beyond the roast profile itself: batch-size consistency, green coffee temperature at charge, ambient temperature and humidity (which affect the drum environment), and bean-density consistency between bags of the same lot.
High-volume specialty roasters invest heavily in consistent green coffee storage conditions, standardised pre-roast protocols, and sensor calibration to make replication as reliable as possible. Without this infrastructure, even a perfect profile on paper will produce variable results batch to batch.
A roast curve is both a plan and a record. The plan is what the roaster intends. The record reveals what actually happened. The gap between them is where roasting skill is built.
Sensory Feedback: Data and the Cup
Roast curve data is a tool, not an endpoint. Every profile must be evaluated by cupping the result, and the cupping informs adjustments to the profile for the next roast.
Experienced roasters use a two-stage feedback loop: real-time data monitoring, tracking RoR, bean temperature, and key event timing during the roast; and post-roast cupping, tasting the result to assess development, balance, acidity, sweetness, and origin character, and making profile adjustments based on what the cup reveals.
Neither data alone nor sensory evaluation alone is sufficient. The roaster who only watches the screen is ignoring the most important information. The roaster who only tastes and never monitors is flying blind with repeatable disasters.
The interaction between data and sensory assessment is what separates technically informed specialty roasting from craft roasting without feedback.
