The 7 Stages of Bean-to-Bar Production Explained
A stage-by-stage field guide to bean-to-bar chocolate production — sorting, roasting, winnowing, grinding, conching, tempering, and molding. Temperatures, timings, failure modes, and the flavor consequences of getting each step wrong.
Every bean-to-bar bar in the world — from a $4 grocery-co-op house bar to a $22 single-estate Chuao — passes through the same seven production stages. Each stage has a specific purpose, a set of tolerances, and a failure mode that will quietly degrade the finished bar if you get it wrong. This is a working field guide to all seven, written for makers who want to understand what each step is actually doing to the chocolate in their hands.
If you're new to the category, start with our complete guide to bean-to-bar chocolate making for the business and equipment context. If you want to understand the pre-production side of things — the four to seven days that happen before a bean ever reaches you — read our cacao fermentation primer. This post picks up the moment a sack of fermented, dried beans lands on your receiving bench.
Stage 1 — Sourcing & sorting
The production journey starts before the roaster turns on. Every sack of cacao arrives with some amount of foreign matter — twigs, stones, flat beans, moldy beans, pieces of bean husk from earlier processing. A sacred rule of bean-to-bar chocolate making: the bar can only be as clean as the sort. A single moldy bean in a 25-kg sack can throw an off-note across an entire 55-bar batch.
What a good sort looks like
- Spread beans in a single layer on a clean tray, roughly 1 kg at a time.
- Remove: any bean that is flat (unfilled), moldy (whitish or greenish surface), germinated (sprout visible), heavily cracked, or noticeably lighter than the rest of the lot.
- Pull any non-bean debris — twigs, stones, bits of fiber.
- Target a removal rate of 3–6% by weight. If you're removing less than 2%, you aren't looking carefully. If you're removing more than 8%, the lot may be too poor to use.
Stage 2 — Roasting
Roasting develops flavor. Specifically, it drives Maillard reactions between the free amino acids and reducing sugars created during fermentation, generating most of the volatile aroma compounds that make chocolate smell like chocolate. It also drives off residual acetic acid from fermentation, reduces bean moisture to workable levels, and loosens the husk so it can be winnowed.
Target ranges
| Variable | Typical range | Notes |
|---|---|---|
| Bean temperature | 110–140 °C | Measured at the bean mass, not air temp |
| Duration | 12–35 min | Shorter for delicate origins, longer for robust |
| Rate of rise | ~2–3 °C/min mid-roast | Gentle curve, not a ramp-then-flat |
| End moisture | 1.5–2.5% | Down from 6.5–7.5% raw |
What “under-roasted” tastes like
Sharp acidity, green notes, raw nut character, papery astringency. The Maillard reactions haven't had time to develop. The bar tastes closer to raw cacao than to chocolate.
What “over-roasted” tastes like
Flat, generic “chocolate flavor” with no origin character. Burnt, ashy, coffee-like notes. Over-roasting destroys the delicate top-note esters that make a Madagascar taste like Madagascar instead of like a Hershey bar.
The first roast was flat. The second was sharp. The third was the one the bean actually wanted — I could taste the plum and the brown-butter that were hiding the whole time.
Stage 3 — Winnowing
Winnowing cracks the roasted beans and separates the papery husk (shell) from the cocoa nib. The husk makes up roughly 11–14% of roasted bean weight; cocoa butter content and flavor live in the nib, not the husk.
Yield is the number that matters
A well-tuned winnower produces 88–92% nib yield with minimal nib in the husk reject. A poor setup can drop that to 78% — meaning every 100 kg of roasted beans becomes 78 kg of usable nib instead of 90 kg. That 12-point yield swing is worth more than $0.10 per finished bar and accounts for a surprising fraction of cost variance between well-run and badly-run makers. We break down the economics in our true cost-per-bar guide.
Tuning a winnower
- Airflow too high — nib gets blown into the husk catch. Yield loss.
- Airflow too low — husk stays in the nib. Downstream flavor contamination; melanger wear.
- Roller gap too wide — beans pass through whole. No separation.
- Roller gap too tight — nibs get crushed into powder that blows into the husk reject.
Most winnowers need 15–30 minutes of tuning per origin because bean size and density vary. Lots from the same farm but different harvests can behave differently.
Stage 4 — Grinding & refining
Grinding reduces the nib (and added sugar, and often cocoa butter) to a particle size small enough that the tongue can't detect grittiness — below roughly 20 microns. For a small maker, grinding and refining happen in the same piece of equipment: a stone melanger running 24–72 hours per batch.
How particle size affects mouthfeel
| Particle size | Mouthfeel |
|---|---|
| >40 microns | Obviously gritty. Unfinished. |
| 25–40 microns | Detectable grit, especially on the finish. |
| 18–25 microns | Smooth but with distinct texture. Target for rustic bars. |
| 12–18 microns | Silky. Target for polished dark and milk. |
| <12 microns | Very smooth but can feel flat or slippery. |
A stone melanger running at ~40–45 °C for 36–48 hours typically brings most recipes to 15–18 microns. Longer run times don't necessarily mean finer particles — the wheels and stone reach a steady state — but they do affect conching, which is the next stage.
Stage 5 — Conching
Conching is extended mechanical agitation of the liquid chocolate — often at slightly elevated temperatures — that does three things the grinder alone can't. It drives off remaining volatile acids (especially residual acetic acid from fermentation), it smooths mouthfeel by fully coating particles in cocoa butter, and it develops rounded, integrated flavor by giving volatile compounds time to evaporate and interact.
Small bean-to-bar makers usually conche inside the same melanger that ground the chocolate — the stone wheels keep turning for another 24–48 hours after particle size has stabilized. Larger makers have dedicated conches that can run hotter and drier.
How conching changes flavor
- Under-conched(under ~18 hours of total melanger time): sharp acidity, volatile harshness, edges you'd describe as “young.”
- Well-conched (36–72 hours total): round, integrated, mouthfeel is silky, finish is long.
- Over-conched(90+ hours): delicate aromatics evaporate. Bar tastes “polished” but loses its origin character. Industrial chocolate is often deliberately over-conched to produce uniform flavor at scale.
Stage 6 — Tempering
Tempering is the most mechanical, most failure-prone step in the whole chain. It's what coaxes the cocoa butter in your chocolate into the stable Form V crystal structure — the one that gives the finished bar a clean snap, a glossy surface, a clean release from the mold, and shelf stability against cocoa butter bloom.
The three-stage tempering curve
- Melt completely — bring the chocolate to 45–50 °C to erase any existing crystal structure.
- Cool to seed temperature — drop to 27–28 °C (dark) or 26–27 °C (milk) to encourage Form V formation.
- Warm slightly for working — bring back up to 31–32 °C (dark) or 29–30 °C (milk) to melt out any lower-form crystals while keeping the Form V seed.
Tempering methods at small scale
- Seeding — adding a small amount of already-tempered chocolate as a crystal seed while the batch cools. Low equipment requirement, high skill requirement.
- Tabling — pouring chocolate onto a marble slab and agitating it with a spatula until crystals form. Classic, mesmerizing to watch, tiring for batches over a couple of kilograms.
- Machine tempering — a continuous temperer (ChocoVision Rev, Selmi, Kreuzer) handles the curve automatically. The default for any maker producing more than a few hundred bars a week.
An out-of-temper bar shows up as one or more of: dull surface, crumbly or snapless break, sticky release from the mold, or bloom (a whitish, dusty-looking coating that develops over days or weeks of storage). Tempering rejects are typically 3–5% of a solo maker's batch — a cost line we cover in the cost-per-bar guide.
Stage 7 — Molding, wrapping & aging
The final stage converts tempered chocolate into finished, saleable bars. The steps look simple but each one is a place where bars get broken, mis-weighed, or cosmetically damaged.
Molding
Pour tempered chocolate into polycarbonate molds, vibrate or tap to release air bubbles, scrape the surface flat, and cool. Most solo makers cool at room temperature (18–20 °C) rather than in a refrigerator — too-rapid cooling can cause unstable crystal forms to set before Form V has time to propagate.
Demolding & QA
A well-tempered bar pops cleanly out of the mold with a glossy surface. Any bar with visible bloom, air pockets on the face, or a sticky release gets pulled to the reject pile. Pull 1–2 bars from every batch for sensory QA — break, snap, gloss, mouthfeel, flavor.
Wrapping
Foil inner, printed outer wrap, label or sticker seal. Every touch adds risk of fingerprints and warmth damage — most small makers wrap in a climate-controlled room below 20 °C. Thin cotton gloves are cheap insurance.
Aging
Many serious makers age finished bars for 2–8 weeks before release. Aged bars taste noticeably rounder — remaining volatile acids continue to dissipate, and the fat crystal structure completes its final organization. If you've never done a side-by-side taste of a day-1 bar vs. the same bar four weeks later, do one. It's the cheapest quality improvement in your process.
How the stages interact
The hardest thing about bean-to-bar chocolate is that the seven stages aren't independent — each one shifts what the next needs. Under-ferment the beans and you'll need a longer, hotter roast to correct; under-roast and you'll need a longer conche to drive off the acid; conche too long and you'll flatten the flavor the roast just developed. Mastery isn't about memorizing temperatures — it's about tasting the bean at every stage and adjusting the next step to fit what the bean actually needs.
The first three years I was executing a recipe. The next three years I was executing the bean. The bar stopped being something I was trying to make and started being something I was trying to hear.
A cheat-sheet for the seven stages
| Stage | Key output | Failure mode |
|---|---|---|
| 1. Sort | Clean, uniform beans | Off-notes from moldy / defective beans |
| 2. Roast | Developed flavor precursors | Under: sharp, raw. Over: flat, burnt. |
| 3. Winnow | 88–92% nib yield | Husk in nib; yield loss |
| 4. Grind | <20 µm particle size | Gritty mouthfeel |
| 5. Conche | Rounded, integrated flavor | Under: harsh. Over: lifeless. |
| 6. Temper | Form V crystal structure | Bloom, dull surface, poor snap |
| 7. Mold / age | Finished, stable bar | Air pockets, tempering damage, premature release |
If you want the business side of the operation that sits around these seven stages — recipes, batches, true cost, wholesale, and compliance — read our complete bean-to-bar guide. And if you're trying to decide how to structure your catalog around what you've learned about origin and blend behavior, our single-origin vs. blend analysis lays out the framework. Welcome to the craft.