When roasting coffee, how the coffee is roasted has a large effect on the final flavor of the coffee. Every lot of coffee requires its own unique roast profile to bring out the desired character of that coffee. Even when roasting the same mark of coffee from the same farm to get the same flavor from one season to the next may require changes to how that coffee is roasted. In order to achieve the desired results, there must be a procedure for determining the desired roast profile. The following procedure works for me. It may work for you, but it won’t work for everybody. To use this procedure, you need a coffee roaster that is both controllable and provides good bean temperature measurements at batch sizes small enough that you won’t mind wasting a few batches figuring a out a profile. The roaster should also have a well placed and sufficiently large sample trowel to extract a usable sample of coffee quickly during the roast.

Start by roasting the coffee in several ways. The easiest way to do this is to roast a small batch of coffee and pull several samples from that batch. These samples should cover a range from too light through too dark with as many samples in between as possible. What degree of roast is too light or dark is a judgement call based on what you’re looking for in that coffee (perhaps informed by an earlier sample evaluation) and personal or customer preferences. It may be useful to do this with a few small batches to explore the effect of different airflow manipulation patterns or different amounts of time in important regions of the roast. If you’re not in a hurry to figure out the roast profile and already understand how these changes influence the flavor of the coffee, start with one batch and use the evaluation of the first set of samples to guide further batches if required.

When roasting these batches, it is important to remember that the final goal is a reproducible roast profile. This means that you need to understand the limits of your roaster with larger batch sizes to avoid designing a profile that is impossible to duplicate with a larger batch. It also means that it is important to keep a record of the roast profile and the time and bean temperature at which each sample was pulled. Using some sort of automated data logging such as Typica (which was designed with this sort of use in mind) is helpful but not required. If you opt for manual data logging and are uncoordinated, get another person to record the data as you roast and pull samples.
After pulling these samples, the coffee needs to be cooled. Due to the small amount of coffee in each sample, little is needed to cool the sample in an amount of time similar to the cooling time for a full batch. I collect my samples in small glass cups and shake these in such a way as to toss the coffee into the surrounding cooler air.
The samples should rest at least four hours before they are evaluated. Shorter resting times can be used with practice, but the flavor of these samples will not match as well with how anybody else will drink the coffee. I have not determined an upper limit on the amount of resting time, but letting the samples rest overnight is fine as long as the samples are kept free of environmental contamination.
Cupping for Roast Profile Development
Once the roasted samples have rested, it is time to cup them. The procedure is similar to that used to evaluate green coffees, but there are some important differences. The amount of coffee available from each sample is much smaller. You’ll only have one cup from each sample with a small amount left over. You’ll also often be working with a large number of distinct samples which makes most cupping forms unwieldy. I use 12 grams of coffee ground to setting 5 on a freshly calibrated Ditting grinder in ceramic cups with a capacity of 8 fluid ounces. Don’t trust those numbers. Run your own experiments to figure out what cupping procedure works best for you.

You don’t need to rank these samples. Remember that the goal is to determine what roast profile you want to use. You will be rejecting most of the samples very quickly and no observations need be noted for those samples that cannot be used. Hopefully, you find a sample that matches what you want out of that coffee. If you do, you can move on to the next step. Otherwise, you’ll need to roast another set of samples using your observations and experience to determine what changes to the roast profile are most likely to produce better results. You may want to write a brief description of the sample you’ve chosen. If you find a sample that you have no immediate need for but may find useful later (for example, in blending), it can also be useful to make a note of that, but standard evaluation forms are a poor fit at this point.

Once the desired roast profile is found, a batch of coffee should be roasted to match that profile. This is done to ensure that the chosen roast profile can be matched. An evaluation of this coffee will further verify that this is really the roast profile you want. I’ll usually prepare a drip pot of the coffee as that is how most people will be enjoying it. This evaluation can be used to produce a marketing description of the coffee and to set the quality assurance parameters future batches will be judged against.
