Modern coffee roasting generates an overwhelming amount of data. Temperature curves, rate of rise (RoR), delta time, energy application – the numbers are precise, objective, and relentless. But numbers alone do not make great coffee. The ultimate test is always flavour. The challenge is learning to translate data points into predictable taste outcomes. This guide bridges the gap between what you see on your Artisan Scope screen and what you taste in the cup.

The Problem with Data-Only Roasting

Many roasters fall into one of two traps. The first trap is ignoring data entirely and roasting by sight, sound, and instinct alone. This approach can produce excellent coffee but is nearly impossible to replicate consistently. The second trap is worshipping data, chasing perfect curves while ignoring sensory feedback. This produces mathematically beautiful roasts that taste hollow or flawed.

The solution is integration. Data tells you what happened. Cupping tells you what it means. By systematically connecting specific curve shapes to specific flavour outcomes, you build a translation layer between numbers and taste.

The Essential Metrics That Matter

Before connecting metrics to flavour, you must understand what each metric actually measures.

Bean Temperature (BT) is the physical temperature of the bean mass. It is the most intuitive metric but the least useful in isolation. Two roasts with identical end temperatures can taste completely different based on how they arrived there.

Rate of Rise (RoR) is the derivative of bean temperature, measured in degrees Celsius per minute. This is the most important metric for flavour control. RoR tells you how fast heat is being absorbed. A declining RoR curve is the signature of a well-developed roast. A flat or rising RoR curve indicates imminent baking or scorching.

Environmental Temperature (ET) measures air temperature inside the drum. The gap between ET and BT represents thermal momentum. A large gap indicates aggressive heat application. A narrow gap indicates gentle, extended development.

Delta Time tracks the duration of each phase: drying (charge to yellowing), Maillard (yellowing to first crack), and development (first crack to drop). The ratio between phases determines the balance of sweetness, acidity, and body.

Phase One: Drying (Charge to Yellowing)

The Data Signature

During drying, surface moisture evaporates and beans turn from green to yellow. Target duration is typically 4 to 6 minutes for a 300g batch. RoR during this phase should be high and steadily declining, starting around 15 to 20°C per minute and falling to 10 to 12°C per minute by yellowing.

Connecting to Flavour

Too fast (under 3 minutes): The exterior of the bean heats too quickly before the interior can conduct heat. This produces a sharp, vegetal acidity and a thin body. Grassy and peanut flavours dominate.

Too slow (over 7 minutes): The beans lose too much moisture before the Maillard phase begins. The result is a baked flavour profile with muted acidity and a shortened finish. Sweetness is lost.

Just right (4 to 6 minutes): The bean structure is preserved. Sugars are intact. The coffee will have the potential for balanced acidity and a clean finish.

The Adjustment Rule

If drying phase is too short, lower charge temperature or reduce heat application. If drying phase is too long, raise charge temperature or increase initial heat. A change of 5°C charge temperature shifts drying time by approximately 30 seconds.

Phase Two: Maillard (Yellowing to First Crack)

The Data Signature

The Maillard phase is where amino acids and reducing sugars react to create flavour precursors. Target duration is 3 to 5 minutes. RoR should decline smoothly from 10°C per minute at yellowing to 5 to 6°C per minute at first crack. The curve should be a straight or slightly concave line, never convex or flat.

Connecting to Flavour

Too fast (under 3 minutes): The chemical reactions are interrupted. The coffee will taste underdeveloped despite hitting first crack. Acidity is aggressive and sour rather than bright. Sweetness is minimal.

Too slow (over 5 minutes): The roast stalls thermally, even if RoR never reaches zero. The coffee tastes baked and flat. Acidity is completely muted. The finish is short and grainy.

A flat RoR line during Maillard: This is a subtle stall. The coffee will taste hollow with a papery mouthfeel. Even experienced roasters often miss this because the end colour looks correct.

A rising RoR line during Maillard: This indicates increasing thermal momentum, which almost always leads to scorching or tipping. The flavour will have sharp, burnt notes.

Just right (3 to 5 minutes with declining RoR): The coffee develops clean acidity, perceptible sweetness, and a medium body. Floral, fruity, and chocolate notes become distinguishable.

The Adjustment Rule

To extend Maillard, reduce heat before yellowing or increase airflow. To shorten Maillard, increase heat or reduce airflow. Never change charge temperature to adjust Maillard alone, as this affects drying phase simultaneously. Adjust one variable at a time.

Phase Three: Development (First Crack to Drop)

The Data Signature

Development begins at first crack, an audible and thermal event visible on the RoR curve as a flick or brief plateau. Target development time ratio (DTR) is typically 15 to 25 percent of total roast time. RoR should continue declining to 2 to 4°C per minute at drop.

Connecting to Flavour

Under 15 percent DTR: The coffee is underdeveloped even if colour looks correct. Expect vegetal, grain-like flavours. Acidity is sharp and unpleasant. No sweetness is present.

15 to 18 percent DTR: Bright acidity with minimal body. Suitable for high-quality washed Africans where clarity and fruit notes are the goal. Sweetness is present but light.

18 to 22 percent DTR: Balanced acidity, body, and sweetness. This is the sweet spot for most washed Centrals and Ethiopians. Origin character is preserved.

22 to 25 percent DTR: Reduced acidity, increased body, deeper sweetness. Suitable for Brazils, Colombians, and espresso blends where chocolate and nut notes are desired.

Over 25 percent DTR: Acidity is gone. Roast flavours dominate. The coffee will taste bitter and ashy, suitable only for very dark roasts or specific traditional profiles.

The Adjustment Rule

Development time is controlled by heat application during Maillard and early development. To increase DTR without entering second crack, reduce heat gradually as first crack approaches. To decrease DTR, maintain higher heat through first crack and drop immediately after cracks subside.

Case Study: Two Roasts, Same End Temperature, Different Flavour

A practical example demonstrates why data matters. Two 300g batches of a washed Guatemalan are roasted to the same end temperature of 210°C.

Roast A: Drying 4:00, Maillard 4:30, DTR 18 percent. RoR declines smoothly from 18 to 4°C per minute. The curve is a clean downward slope.

Roast B: Drying 3:00, Maillard 5:30, DTR 22 percent. RoR is erratic, with a flat section during Maillard and a brief rise at first crack.

Both roasts look similar to the naked eye. Both have identical end temperatures. But Roast A tastes of chocolate, apple, and clean caramel. Roast B tastes baked, flat, and slightly astringent. The data explains why. The flat RoR in Roast B indicates a stall, even though total development time is longer. The longer development did not save it because the thermal curve was broken.

Building Your Own Flavour Translation Chart

For each green coffee you roast, create a chart that maps data ranges to taste outcomes. Include the following columns: green coffee origin and processing method, optimal drying time range, optimal Maillard time range, optimal DTR percent, target RoR at yellowing, target RoR at first crack, target RoR at drop, expected flavour notes when on curve, and flavour defects when off curve.

After each roast, cup the coffee blind. Write down flavour notes. Then look at the data curve. Ask specific questions. Did the drying phase run long or short? Did RoR decline smoothly or flatten? Was DTR within target? Over time, patterns emerge. You learn that a certain curve shape consistently produces berry notes. Another curve shape always tastes nutty. You no longer guess. You predict.

The Limits of Data

Data cannot taste coffee. Data cannot tell you that a slightly longer development makes this particular lot of Kenyan taste better even though the numbers say otherwise. Data is a tool, not a master. The best roasters use data to guide decisions, then confirm with their senses.

When a roast tastes excellent despite an unusual curve, save that curve and investigate. When a mathematically perfect curve tastes bad, trust your palate and adjust. The goal is not to achieve the most beautiful Artisan screenshot. The goal is to produce coffee that makes people want another cup. Data is the map. Flavour is the destination. Learn to read both.