Your cart is empty
Browse OfferingsLicensed & Affiliated
Ethio Coffee Import and Export PLC is a family-owned Ethiopian coffee exporter shipping green coffee beans to roasters, importers, and distributors worldwide.
© 2026 Ethio Coffee Import and Export PLC. All rights reserved.
Reference frameworks:
A buyer's sample roast is a measuring instrument, not a miniature production roast. Define one purpose, use a stable machine state and fixed batch size, aim for the same verified roast level, cool and label every batch identically, then reject the roast before judging the coffee if scorching, tipping, uneven development, or a missed endpoint could distort the cup. Roast the offer, pre-shipment, arrival, and control samples under the same written protocol. Record enough data for another operator to reproduce the result and for procurement to explain the final lot decision.
A green coffee sample roasting protocol protects a buyer from a quiet but expensive error: blaming the lot for a bad roast. A scorched offer sample can hide floral clarity. An underdeveloped pre-shipment sample can create false cereal or peanut notes. A darker arrival roast can make a sound shipment appear different from its retained reference. If preparation changes at each checkpoint, the resulting scores cannot support a confident purchase, release, or claim.
The remedy is not one universal temperature curve. Sample roasters transfer heat differently, probe readings are not interchangeable, and batch sizes change machine response. A useful protocol controls the conditions that can be repeated in your lab, verifies the final roast, and separates roast failure from coffee failure. The procedure below is designed for importers, roasters, exporters, traders, and quality teams evaluating commercial green coffee lots.
Start by writing the decision above the roast log. An evaluation roast asks whether the green coffee's inherent character, consistency, and defects meet a buying specification. A production-development roast asks how to turn an already selected coffee into a product. Combining those jobs introduces preference before procurement has established baseline quality.
The International Trade Centre describes sample roasting as a tool for quality and uniformity assessment, product development, and sales or education, while stressing consistency as the primary goal. Those uses are legitimate, but they should be separate batches with separate labels. Approve or reject the green coffee from a neutral, repeatable evaluation roast. Explore filter, espresso, or blend potential only after that gate.
| Roast job | Question it answers | Allowed adjustment | Commercial use |
|---|---|---|---|
| Evaluation roast | What quality is present in the green coffee? | Only the bounded changes needed to hit the house endpoint cleanly | Selection, PSS approval, arrival release, inventory QC |
| Development roast | How could this coffee perform in a chosen product? | Profile changes guided by the intended brew, roast style, and customer | Menu design, blend work, production transfer |
| Control roast | Is the machine and operator reproducing the lab standard today? | None until the deviation is diagnosed and documented | Session release and operator calibration |
Do not force a small offer sample to answer every question. Reserve enough green coffee for physical inspection, one evaluation roast, a repeat if that roast fails, and a product-development roast where needed. If quantity is tight, protect the repeatable evaluation first and request more coffee before making a production claim.
A perfect roast of the wrong sample proves nothing. Link every bag or jar to the exporter, contract, lot, crop year, origin, process, grade, sample stage, sampling date, and date received. Assign a blind code for the cupping table without removing the traceable laboratory ID. Photograph the sample and label before splitting it.
The buyer's three commercial checkpoints are the offer sample, pre-shipment sample, and arrival sample. The Ethiopian coffee sample approval workflow explains what each checkpoint authorizes. Sample roasting supplies comparable sensory evidence inside that workflow; it does not establish whether a pre-shipment sample was drawn representatively or whether an arrival sample belongs to the shipment.
Moisture and density help an operator anticipate heat demand, but they are not permission to give favored lots an easier roast. Use the measurement methods in the green coffee moisture and water activity guide and green coffee density QC protocol, then keep all adjustments inside the written house limits.
Current SCA standards list SCA-102 Coffee Value Assessment: Sample Preparation and Tasting Mechanics as the preparation framework. Its sample-roasting principle is outcome-based: reach the intended roast level, avoid roasting problems, and keep conditions as even as possible across samples. The SCA's published preparation materials identify an Agtron Gourmet ground reading of 63 as the cupping roast target. Labs using another meter should document the corresponding scale and validation, not assume the displayed numbers are interchangeable.
That distinction matters because the familiar eight-to-twelve-minute rule comes from the heritage cupping protocol and remains a useful diagnostic window on many drum machines, but it is not a universal recipe for every current sample-roasting technology. Treat time as a controlled house parameter after calibration. Treat verified endpoint and freedom from roast defects as the acceptance criteria.
ISO 6668 takes a broader trade approach. It specifies preparing green coffee for sensory analysis and notes that consistent preparation may support comparative assessment or shipment acceptance when parties agree. A buyer can therefore reference SCA-102 for the lab method, while the contract separately names the approved lab, reference sample, tolerances, response deadline, and commercial remedy.
| House-standard field | What must be fixed or verified | Why it matters |
|---|---|---|
| Machine and batch | Roaster ID, software or firmware version, batch mass, preheat state | Defines the thermal system being repeated |
| Environment | Room temperature, significant humidity or pressure change, exhaust condition | Explains session drift and altitude effects |
| Profile controls | Charge rule, power and airflow steps, target time band, drop rule | Lets another trained operator repeat the roast |
| Endpoint | Validated color scale, sample preparation, reading window, tolerance | Stops time or probe temperature from standing in for development |
| Cooling and storage | Cooling method and maximum time, container, roast and cup timestamps | Prevents continued roasting and unequal post-roast handling |
| Acceptance | Color tolerance, visual faults, control-coffee tolerance, repeat authority | Keeps a failed roast from becoming a failed lot |
Build the baseline with a stable control coffee before evaluating live offers. Choose a clean, well-known lot with enough retained volume to cover several calibration sessions. Bring the roaster to thermal equilibrium, not merely to the displayed charge temperature. On a drum machine, one or more seasoning batches may be necessary. On an electric profile roaster, follow the manufacturer's warm-up and between-batch recovery procedure.
Do not copy another laboratory's charge temperature into your SOP. Probe placement, sensor type, drum material, airflow, burner power, and batch-to-capacity ratio can make identical displayed temperatures produce different coffee. Calibrate the profile by repeated control roasts, verified ground color, physical inspection, and blind sensory results. Record the machine-specific settings as your operating method, not as an industry truth.
Mass loss can flag drift but should not replace color or tasting. Two coffees with different initial moisture and composition can reach the intended sensory endpoint at different mass-loss percentages. Use it as a trend line within comparable coffees. Investigate a sudden change before judging the lot.
A familiar curve is evidence only on the machine that produced it. If a dense or high-moisture coffee moves differently, the operator may use pre-approved bounded corrections to reach the shared endpoint. Freehand changes that make every coffee follow the same screen trace can introduce scorching, stalling, or excess development while the graph looks tidy.
Ethiopian samples can differ in density, screen distribution, moisture, processing style, and physical uniformity even within one region. A washed high-elevation Yirgacheffe may absorb heat differently from a natural Sidama lot. Small screen beans can color externally before larger beans in a mixed sample. Naturals may also reveal quakers only after roasting. The protocol should anticipate those differences without rewarding or penalizing an origin through a different target.
| Observed condition | Controlled response | Do not do this |
|---|---|---|
| Higher density or moisture | Use the validated higher-energy or recovery variant, then verify the same endpoint | Extend the roast indefinitely until the outside looks dark enough |
| Broad screen distribution | Record the distribution, use the approved airflow strategy, inspect internal and external evenness | Remove small beans unless the contract sample is specified by screen |
| Natural-process lot | Watch rate near first crack, cool promptly, count pale quakers after roast | Darken the batch to conceal uneven color or fermentation character |
| Low sample quantity | Use the validated small-batch profile or request more sample | Run an unvalidated partial charge and treat the score as contractual evidence |
Limit the house protocol to a few validated profile families, such as standard, higher-energy, and small-batch. Define the objective trigger for each one. The operator chooses from that library before the roast based on measured green properties, not after seeing the supplier name or expected score. The endpoint, cooling rule, roast-fault criteria, and cupping method stay constant.
Origin-side and destination labs should exchange more than scores. Share the roaster model, batch size, profile version, ground-color result, roast date, cup date, and green measurements. That data will not erase differences in water, altitude, equipment, or assessors, but it makes a two-point score gap diagnosable instead of personal.
Give the roast its own pass or fail decision. The operator records that decision before cuppers see a score. A roast outside the color tolerance, visibly scorched or tipped, badly uneven without a green explanation, or affected by a machine interruption should be repeated from the reserved sample. Mark both records; never delete the failed batch from the audit trail.
Target endpoint is within tolerance, development is even enough for the sample type, no material scorch or tip is present, cooling and labeling are correct, and the control coffee is in range.
One indicator is unusual but not clearly disqualifying. Hold the batch, inspect ground color and internal development, review the control, and let the authorized quality lead decide whether to repeat.
Endpoint is missed, roast faults are visible or sensory contamination is likely, the machine alarmed, cooling failed, identity is uncertain, or the control coffee shows that the session is out of calibration.
| Possible roast problem | Physical or cup signal | Buyer action |
|---|---|---|
| Underdevelopment | Very light interior, cereal, raw nut, grassy sharpness | Repeat before attributing the note to the green coffee |
| Scorching or tipping | Localized dark marks, smoky bitterness, harsh finish | Reject the roast and inspect charge energy and agitation |
| Stalling or baking | Muted aroma, flat sweetness, papery or hollow presentation | Repeat and investigate energy or airflow through the middle and late roast |
| Excess development | Roast flavor, bitterness, loss of origin clarity | Repeat to the target endpoint; do not lower the coffee score |
| Green-coffee nonuniformity | Pale quakers or mixed development persists across valid repeats | Record the physical evidence and assess it against the lot specification |
One repeat answers whether preparation caused the result. Repeatedly roasting until a preferred score appears creates selection bias. The SOP should allow a repeat for a defined roast failure, identity problem, or authorized investigation, while preserving every result and naming which batch supports the commercial decision.
Seal accepted batches in clean, odor-free containers and use the same post-roast interval for all coffees in the session. Record roast and cup times. Grind and brew with the adopted sensory standard, consistent water, dose, vessel volume, grind distribution, and sequence. The Ethiopian coffee cupping and evaluation guide covers the sensory stage in detail.
Roast the retained reference and candidate sample in the same session whenever the decision depends on a match. Randomize cup positions and use blind codes. Place a stable control coffee in the session. If the control cups outside its established range, investigate water, grinder, roast, and panel calibration before approving or rejecting a commercial lot.
Small sensory movement does not automatically mean contract failure. Crop samples are biological materials, and the offer-to-PSS agreement should define tolerances for score, cup profile, defect, moisture, screen, and other material attributes. Where a significant mismatch remains after valid preparation, document it through the green coffee quality claims process instead of debating roast curves by email.
The log should let a second operator reproduce the roast, a quality manager audit it, and a buyer connect it to the contract decision. Keep raw values as well as pass or fail labels. If software stores the curve, export or back up the record under a stable sample ID rather than relying on a private operator account.
| Log group | Required fields |
|---|---|
| Identity | Lab ID, blind code, supplier, lot, contract, sample stage, crop year, origin, process, grade |
| Green condition | Date received, package condition, moisture, water activity, density, screen, defects, green mass |
| Roaster state | Machine ID, profile version, operator, date, ambient notes, preheat and seasoning status |
| Roast events | Charge, turning point if meaningful, yellow, first crack, drop, power and airflow changes, alarms |
| Verification | Roasted mass, mass loss, ground color and scale, cooling time, physical faults, roast acceptance |
| Sensory and decision | Cup time, method, control result, score or assessment, match status, decision, approver, response date |
Version the profile whenever a meaningful control changes. A replacement probe, software update, exhaust service, different batch size, new color meter, or relocation may require revalidation. Keep the old version attached to historical roasts. Otherwise a future quality investigation may compare results produced by two procedures that happen to share one name.
A small roastery can implement this system without a multi-barrel lab. The green coffee quality control lab setup guide describes equipment tiers, including a logged small-batch or production-roaster approach at low sample volume. The essential controls are stable batch preparation, a repeatable profile, a verified endpoint, a reference coffee, and documented acceptance.
Audit the process against decisions, not paperwork volume. A complete log that cannot identify which roast was cupped or which lot was approved is still a failed record. Conversely, a concise record that preserves identity, method, endpoint, acceptance, sensory result, and authority can resolve questions months after the coffee enters production.
A green coffee sample roast prepares a small, representative batch for consistent sensory evaluation. Its purpose is to reveal inherent quality, uniformity, and defects without letting roast character dominate. Buyers use accepted sample roasts to evaluate offers, approve pre-shipment samples, verify arrivals, and monitor stored lots.
SCA preparation materials identify a target ground reading of 63 on the Agtron Gourmet scale for the standard cupping roast. Record the meter, scale, sample preparation, measurement timing, and house tolerance. Readings from different instruments are not automatically interchangeable, so validate the corresponding target before use.
No universal time fits every machine. Eight to twelve minutes is a familiar heritage-protocol range and a useful starting diagnostic for many drum roasters. Current SCA guidance emphasizes the intended roast level, absence of roasting problems, and consistent conditions. Validate a machine-specific time band instead of copying temperatures or times.
Yes, if the machine can roast the available sample mass repeatably and the lab validates a stable profile and endpoint. Production schedules and low batch-to-capacity ratios often reduce consistency. When sample volume grows, a dedicated sample roaster usually improves turnaround, repeatability, and separation from production contamination.
Repeat when the roast misses its endpoint, shows scorching, tipping, stalling, uneven development beyond the sample's condition, machine interruption, failed cooling, or uncertain identity. Preserve the failed record and state why a repeat was authorized. Do not keep roasting until a preferred cup score appears.
A green coffee sample roasting protocol turns roasting from an undocumented influence into controlled buyer evidence. Fix what your lab can repeat, verify the roast before judging the coffee, use retained references and a session control, and preserve the connection from sample identity to commercial decision. That discipline protects good lots from bad roasts and protects buyers from approving results they cannot reproduce.
Review current lots, request traceable samples, and align offer, pre-shipment, and arrival checks with an origin-connected Ethiopian exporter.