Coffee Roasting Profiles

What a coffee roast profile records

A roast profile is both a record of what happened and a reference for what to do next. Its central graph plots elapsed time horizontally and temperature vertically, with markers for events such as loading the beans, yellowing, first crack and discharge.

A fuller record includes the green coffee identity, batch mass, heat settings, airflow changes and observations. These details help explain why two batches followed different paths rather than merely showing that they did.

The distinction between a profile and a roast level matters. Roast level describes how far roasting has progressed, often assessed through colour; the profile describes the route taken to reach that endpoint. Two batches can finish at similar colours yet taste different because their temperature histories differ.

For the wider process, start with coffee roasting. For the distinction between light, medium and dark, see roast levels.

The phases of roasting

Roasting software commonly divides a batch into three phases. They are useful reporting boundaries, not separate chemical compartments.

“Drying” does not mean that the beans become completely water-free at yellowing. Likewise, Maillard reactions, the browning reactions involving sugars and amino compounds, do not stop when first crack begins.

Development is particularly easy to misread. The term names the final interval in a roast log, but flavour development is not something that suddenly starts at first crack. Earlier heat application remains part of the coffee’s roasting history.

First crack is an audible event used as a timing reference. Mark its start consistently and record uncertainty if the sound is difficult to distinguish. The physical changes and later cracking belong in first and second crack, rather than being inferred from a graph alone.

Reading temperature curves

The main trace is often labelled BT, meaning bean temperature. Treat this as the bean-probe reading, not a direct measurement of every bean’s centre.

Probe exposure matters. A sensor interacting with the moving bean mass can also be influenced by hot air, and changing the batch size can change that exposure. Air-temperature and exhaust-temperature traces measure different locations and should not be mistaken for interchangeable versions of BT.

After loading, the recorded temperature commonly falls before rising again. The minimum is called the turning point. This initial dip reflects the measurement system responding to the incoming cooler coffee; it is not evidence that room-temperature beans first became hot and then cooled.

A labelled diagram description

Read a typical roast graph from left to right:

1. Charge: the beans enter the preheated roaster.
2. Turning point: the recorded bean-temperature curve reaches its early minimum.
3. Yellowing: an event marker ends the logged drying phase.
4. First crack: another marker starts the development interval.
5. Drop: discharge ends the roasting time used for the development ratio.

A second trace, rate of rise, shows the changing slope of the temperature curve. It needs its own scale because degrees Celsius and degrees Celsius per minute are different quantities.

The diagram is a reading guide, not a target shape. Copying another roaster’s temperatures without its machine, probe and batch context is unreliable.

Development Time Ratio

Development time ratio, usually shortened to DTR, expresses the interval from first crack to drop as a percentage of the total roast time.

DTR (%) = development time ÷ total roast time × 100

Consider an illustrative batch with first crack at 8 minutes and drop at 10 minutes. Development lasts 2 minutes, so:

120 seconds ÷ 600 seconds × 100 = 20% DTR

This is a calculation example, not a recommended recipe.

Why the percentage needs context

A shorter illustrative roast could reach first crack at 6 minutes and finish at 7 minutes 30 seconds. Its development interval is only 90 seconds, but its DTR is also 20%.

The same percentage therefore does not mean the same time after first crack. It also says nothing by itself about the temperatures reached during that interval.

Roasting consultant Scott Rao has recommended a DTR of 20% to 25% for particular drum-roasting contexts. His published explanation presents this as an experience-based guideline, acknowledges successful exceptions and explicitly says DTR does not guarantee development. It should not be presented as a universal scientific standard.

Timing judgement adds another limitation. In the 10-minute example, marking first crack 15 seconds earlier changes the calculated DTR from 20% to 22.5%, without changing the roast itself.

Record development duration, total duration and the first-crack marking convention together. The percentage is useful shorthand, not a quality score.

Rate of Rise

Rate of rise, or RoR, measures how quickly the recorded temperature changes. When expressed in degrees Celsius per minute:

Average RoR = temperature change ÷ elapsed time in minutes

For example, a hypothetical rise from 192°C to 198°C over 30 seconds gives an average RoR of 12°C per minute.

A positive RoR means temperature is rising. A falling but still positive RoR means it is rising more slowly, not cooling. Zero means no measured temperature change, while a negative value means the measured temperature is falling.

RoR is mathematically the slope of the temperature curve. It can make changes in heating behaviour easier to notice than the temperature trace alone.

Crashes and flicks

A crash is roaster shorthand for an abrupt fall in RoR, commonly discussed around first crack. It need not cross below zero: a sharp reduction can occur while the measured temperature continues rising.

A flick is a late upward turn in RoR, generally after first crack. It is different from the normal upward movement of the temperature trace.

Some practitioners associate crashes with flat or baked flavours and flicks with stronger roasted character. These are useful working hypotheses to investigate, not diagnoses that a graph can prove. Research on deliberately varied profiles does not establish that every such curve feature produces a particular sensory defect.

If a feature repeats, compare its timing with recorded heat and airflow changes. Avoid applying a universal burner-adjustment schedule to an unfamiliar machine.

Smoothing changes what you see

RoR is calculated from temperature measurements, so measurement noise can appear as spikes. Software uses calculation intervals and smoothing to make the trace readable.

More smoothing can make trends clearer while reducing responsiveness to short changes. A very responsive display may reveal timing more clearly but look much more erratic.

Distinguish the displayed unit from the calculation method. Showing RoR per minute rather than per 30 seconds rescales the value; changing the calculation window or smoothing can change the apparent shape. Keep those settings consistent when comparing batches.

Roast logging software and a useful review method

Artisan and Cropster provide tools for recording roast curves and marking events. Their documentation describes phase displays and profile comparisons; supported data capture and machine-control functions depend on the equipment and connection.

Software is most useful when the record extends beyond the curve. Keep the coffee identity, batch mass, event markers, operating changes and tasting result attached to the same batch.

A practical review sequence is:

1. Check comparability. Confirm the coffee, batch size, machine and measurement settings.
2. Inspect the event markers. Correct obvious logging errors rather than interpreting them as roasting behaviour.
3. Read the whole temperature history. Do not begin with DTR alone.
4. Compare colour and taste. Use the same preparation and assessment method for the batches.
5. Choose one focused change. Record the prediction before the next trial, then check whether the result supports it.

This is a suggested investigation method, not an automatic optimisation routine. The broader judgement involved is covered in the art of roasting.

What the metrics cannot tell you

The strongest reason to log profiles is also the reason to distrust shortcuts: time and temperature interact.

A controlled roasting experiment published in Beverages in 2020 compared development-time changes at similar final roast colour. Shorter development produced fruitier, sweeter and more acidic sensory results, while longer development shifted the tested coffee towards roasted, nutty and bitter characteristics. That supports a profile effect under the study conditions, not a rule that shorter development is always better.

A separate controlled study published in Scientific Reports in 2024 investigated markedly different profiles and measured titratable acidity, a laboratory measure of acid content. Its results cannot establish a universal preference for one RoR shape, because acidity measurements are not a complete sensory assessment.

Use metrics to narrow the question: did this batch heat differently, was first crack marked differently, or did the endpoint change? Then test the explanation in the cup. A repeatable roasting profile is valuable; a repeatably enjoyable result is the purpose.