How a Roaster Profiles a New Coffee: From Sample Roast to the Shelf

There is a number that surprises most people when I tell them, and by “most people” I mean anyone who has not stood in front of a sample roaster at 6 AM wondering whether fourteen seconds more development time will make the caramel pop or bury it. That number is this: before a single bag of a new coffee hits the shelf at Contour Coffee, I have already roasted that coffee somewhere between five and ten times. Every single batch was 100 to 150 grams — roughly enough for one pot of coffee — and every single batch was cupped, analyzed, adjusted, and cupped again. The bag you pick up on a Tuesday afternoon is not the first roast of that coffee. It is not the third. By the time it reaches you, it is the final draft of a process that probably started six weeks earlier with a sample bag and a clipboard full of crossed-out notes.

Profiling is the invisible work of roasting. The customer sees the finished product — the bag, the tasting notes, the roast level label — and none of the trial and error that produced it. That is by design. A good profile should feel inevitable, like the coffee could not possibly have been roasted any other way. Getting there is the opposite of inevitable. Getting there is a process of elimination, informed guesswork, and the occasional moment of sheer discovery that makes fifteen years of Tuesday morning cuppings feel worth it.

The Sample Roast: Where Everything Starts

Every coffee we consider carrying begins as a sample — usually 500 grams of green coffee in a GrainPro bag, shipped from an importer we trust. The sample represents a specific lot: a particular farm, a particular harvest, a particular processing station. It arrives with a spec sheet — altitude, varietal, process, screen size, moisture content — but none of that tells me what it tastes like. For that, I need to roast it.

The sample roast happens on a small machine — ours is a 200-gram capacity drum roaster that looks like a piece of laboratory equipment, which in a sense it is. I will roast 100 grams at a time, roughly enough for three cupping bowls. The roast takes eight to ten minutes, which is faster than a production roast, and the goal is not to make the coffee taste its best. The goal is to make the coffee taste like itself — to develop it enough to evaluate its potential without imposing a roast character that masks the origin. Cuppers call this a “reference roast.” You are not trying to nail the profile on the first pass. You are trying to understand what you are working with.

Twenty-four hours later — coffee needs a full day to degas before cupping, and anyone who tells you otherwise is either lying or selling something — I grind the sample into identical glass bowls, pour hot water directly onto the grounds, wait four minutes, and taste. This first cupping is about big-picture questions: Is the coffee clean? Is it sweet? Is there acidity, and if so, what kind? Does the body have weight or is it thin? Are there defects — grassy underdevelopment, ferment, phenol, anything that should not be there? The first cupping is a gate. If the coffee is not clean and sweet, I do not buy it, and I do not roast it again. Life is too short, and Lakewood has enough good coffee.

The Iterations: Five Roasts, One Coffee, One Clipboard

If the coffee passes the gate, the real work begins. I roast the same coffee five different ways — varying charge temperature, development time, rate of rise, airflow — and cup all five side by side the following morning.

This is where you learn what the coffee wants, as opposed to what you want to impose on it. Some coffees demand a gentle touch — natural Ethiopians, for instance, carry volatile fruit compounds that burn off if you push too hard into first crack, leaving you with a generic “dark roast” and a faint ghost of what could have been blueberry. Other coffees need more heat than you would expect — a dense high-grown washed coffee from Huila can handle development that would flatten a softer bean, and if you under-roast it, you get grassy peanut instead of chocolate and caramel.

When I profiled our current Colombia Excelso EP lot, I roasted it seven times before I was satisfied. Roast one was too light — grassy, vegetal, the citrus tasting more like lemon pith than orange zest. Roast two was a correction: hotter charge, longer development. The grass disappeared, but the chocolate stayed thin. Roasts three and four played with the rate of rise — one faster, one slower. Roast five pulled everything back and tried a different charge temperature entirely. Roast six was close — milk chocolate, caramel, the orange zest finally sitting where it belonged — but the finish was drying and short. Roast seven extended the development phase by twenty seconds and dropped the coffee at 408 degrees instead of 412. Everything locked into place. Milk chocolate on the front. Caramel through the middle. Orange zest bright but gentle at the edges, fading into roasted nut and a clean, sweet finish.

That is what profiling feels like from the inside: seven roasts, seven cupping sessions, one clipboard page of notes where all the entries for roasts one through six are variations on “almost” and “not quite” and “closer, but.” Then roast seven. Then you write “this is it” in the margin and circle the roast curve with a red pen.

The Moment When It Works

There is a specific feeling I have learned to trust over fifteen years. It happens at the cupping table, usually on a Tuesday morning, about thirty seconds into evaluating a new roast. You have tasted the coffee at three different temperatures — hot, warm, and cooling — and at every stage the notes you wanted are present, balanced, and legible. The acidity is bright but not sharp. The body has weight. The finish is clean and persistent. Nothing challenges you. Everything welcomes you.

That feeling is relief, mostly. But it is also satisfaction of a particular kind — the satisfaction of having listened to what the coffee was telling you instead of imposing a preconceived idea of what it should taste like. A roast profile is a collaboration between the roaster and the bean. The bean brings altitude, varietal, processing, and terroir. The roaster brings heat, time, and attention. When the collaboration works, the coffee tastes like itself — fully realized, nothing added, nothing subtracted.

The Ethiopia Guji Natural we carry from Kayon Mountain estate took five roasts to profile. The first roast was too hot and the blueberry went flat — it was there, but barely, a suggestion rather than a statement. The second roast backed off the heat too far and produced grass. The third brought the blueberry back but lost the chocolate entirely. The fourth got the fruit and chocolate in balance but pushed the acidity from bright to sharp. The fifth roast — eleven minutes total, a gentle approach to first crack, slow development, drop temperature 405 — brought it all together. Blueberry forward. Dried strawberry underneath. Lemon brightness. Thyme. Chocolate at the finish. The things that are supposed to be there, exactly where they are supposed to be.

Scaling Up: From 100 Grams to 25 Pounds

The final step is scaling. A roast profile developed on a 200-gram sample roaster does not translate directly to a 25-pound production roaster. The thermodynamics are different — more bean mass means more thermal momentum, different airflow characteristics, different heat transfer patterns. I use software that records temperature curves and allows me to overlay the sample roast curve against the production roast curve, matching rate of rise and development time as closely as the equipment allows. It is never exact. The production roaster has more inertia. Adjustments happen in seconds and degrees, not minutes and tens of degrees.

The first production batch of a new profile gets cupped immediately. If it matches the sample — if the chocolate is where the chocolate was, if the fruit is where the fruit was — the profile is locked and the coffee goes on the shelf. If it does not match, I adjust and roast again. I have roasted a coffee four times at production scale before it matched the sample. That is unpleasant. It is also necessary. Shipping a coffee that does not taste like the profile I approved is a betrayal of both the customer and the farmer. There is no shortcut around this. You roast until it is right, and then you stop.

The next time you open a bag of Colombia Excelso or Ethiopia Guji Natural from Contour Coffee, you are tasting the last roast in a series. The first six or seven roasts of that coffee are gone — consumed at the cupping table, spat into buckets, recorded in a notebook that lives on a shelf in the roastery. What reaches you is the one we kept. The one we decided was ready. Everything that came before it was just practice.

Contour Coffee has been roasting in Lakewood, Colorado since 1979. Every coffee we carry was profiled the hard way — sample by sample, roast by roast, cup by cup — before a single bag reached the shelf. Browse our single-origin lineup and taste the difference that profiling makes.

---

✅ Fact-Checked

Reviewer: Contour Coffee Fact-Checker (mimo-v2.5-pro) Date: 2026-05-20

Verified Claims ✅

• Product references (Colombia Excelso EP, Ethiopia Guji Natural) link to valid catalog pages
• Colombia Excelso tasting notes: “milk chocolate, caramel, orange zest” — matches catalog
• Ethiopia Guji Natural from Kayon Mountain estate — matches catalog
• Sample roasting and cupping methodology — accurate
• “24 hours degas before cupping” — standard practice, accurate
• “Since 1979” brand fact — correct
• “fifteen years” roasting — correct character fact

Concerns ⚠️

• No concerns — educational content with accurate product references.

Changes Needed

• None