Acidity & Brightness
Fresh, fruit-like acidity
- Organic acids · Bright, fresh overall acidity
- Citric acids · Lively citrus brightness
- Malic acids · Apple and pear acidity, vivid in light roasts
The Data · How to read a lot pageExample: lot LG-01 · live data
Read the bean,
not the label.
Every lot on uncrack is decoded before bidding opens, so you bid with evidence in hand.
Here's how to read a lot page, number by number, on a real lot.
Taste potentialPredicted · 0–10
- Acidity
- 8.2
- Aroma
- 3.3
- Flavor
- 6.8
- Sweetness
- 5.3
- Mouthfeel
- 3.0
- Country
- Honduras
- Region
- Francisco Morazán
- Farm
- El Pino
- Variety
- Geisha
- Process
- Washed
- Altitude
- 1700 m
- Harvest
- —
- Lot size
- 35.28 kg
- Moisture
- —
- Density
- —
- TUS uniformity
- 91%
16 componentsThis lotLibrary average
Acidity
Aroma
Sweetness
Body
Bitterness
Stability
Top of lot LG-01's page, live. Hover a bar to see its name.
Origin and specs read as written. The numbers take a closer look, so here are parts 3 to 6, in card order.
Taste potential
A higher score means more, not better.
From the 16 components, the engine predicts five taste scores, each 0–10. Pick the profile that fits your roast.
This lotLibrary average
The shape shows where a lot leans before you ever cup it.
| Taste | A higher score | A lower score |
|---|---|---|
| Acidity | Bright, lively acidity | A soft, rounded cup |
| Aroma | Expressive fragrance, from dry grounds to cup | Restrained; clarity leads, not perfume |
| Flavor | Clean, well-defined flavor | Simpler and more linear |
| Sweetness | Sweetness without pushing the roast | Calmer natural sugars; build them in the roast |
| Mouthfeelalso called Body | Real weight; holds up in espresso and milk | Light body, strongest in filter |
Why isn't bitterness one of the five?
Bitterness depends on the roast, so the engine models it as a roast-dependent potential, not a sixth score. In one line mentions it when it is restrained or bold.
16 components
Sixteen bars, six groups.
Each bar on the card is one component measured in the green bean, scaled 0–1. They're grouped by what they shape in the cup.
Aroma & Flavor Complexity
Aroma, depth and nuance
- Amino acids · Nutty, chocolate, umami
- Diterpene · Fruity, malty complexity
- Trigonelline · Roasts into sweetness, savoriness and aroma
Sweetness Precursors
Sugars that turn sweet on roast
- Sucrose · Primary sweetness; caramelizes in the roast
- Glucose · Clean sweetness; caramel and nutty precursors
- Fructose · Crisp sweetness; berry and tropical aromatics
- Mono sugar · Toasty, rounded sweetness
Mouthfeel & Body
Weight and creamy texture
- Lipids · Creamy texture; carries aroma
- Poly sugar · Body and mouthfeel weight
Bitterness
Cocoa depth, roast structure
- Caffeine · Perceived strength; minor bitterness
- Caffeine sub · Chocolate, cocoa, soft umami
- Chlorogenic acids · Bright when light; smoky-bitter when dark
Stability
Storage and roast reactivity
- Water activity · Roast reactivity and storage stability
The dark tick
Read every bar against its tick.
The dark tick is the library average: every lot we've measured, not only this auction. Above the tick means more than average. Below means less.
One row from the Taste tab · lot LG-01
FlavorThis lotLibrary average6.8above, 19%
above the tick, below it. The % says how far.
- Taste scores
- Strong 12% or more aboveTypical within 12%Restrained 12% or more below
- Components
- More 8% or more aboveSimilar within 8%Less 8% or more below
The same rule on the Chemistry tab
| Component | This | Avg | Diff |
|---|---|---|---|
| Organic acids | 0.48 | 0.35 | above, 38% |
| Mono sugar | 0.81 | 0.64 | above, 28% |
| Chlorogenic acids | 0.75 | 0.59 | above, 26% |
Biggest differences come first, so read the top rows.
Every lot is measured the same way, so you can put any two side by side. In Quality prediction, pick a lot in Compare with. The ticks and percentages switch to that lot.
Labels
Know what kind of number you're reading.
Every number on a lot page carries one of these four labels. Partner cupping shows sample data until panel scores publish.
LAB-MEASURED
Measured
From this lot's own scans. Chemistry and uniformity.
Predicted · 0–10
Predicted
From the chemistry. Not a cupping score. Taste potential.
PREVIEW DATA
Sample values
Shows the format only. Not this lot's result.
—
Not supplied yet
The value will appear once it's provided.
On every lot page
What else is on a lot page.
Seven sections sit below the card, in this order. Open one for a short answer.
01Partner cuppingHas a panel cupped it?Preview panel · sample data
Scores from independent cuppers: the panel average, its low and high, and each cupper. Cupping checks the engine's prediction; it doesn't define it. Until panel scores publish, the section shows the format with sample data.
02In one lineThe profile in one sentence
On lot LG-01: “El Pino Geisha Washed leads with acidity and flavor — a pronounced, expressive profile. Expect a quieter mouthfeel presence.”
03The producerWho grew it?
The producer's name, farm, region and altitude, a photo and story when supplied, and a link to their profile. The name under the lot title opens the same profile.
04Quality predictionHow does it compare, in detail?
Two tabs. Taste: the five scores against the comparison. Chemistry: all 16 components, biggest differences first. Compare with switches the comparison from the library average to a peer lot.
05Quality tierHow good is it likely to be?PREVIEW DATA
The odds that the lot lands in Tier 1, 2 or 3 (preview data until the engine's tier connects). The odds add up to 100%. Ladder and 100 cups show the same odds two ways.
06Uniformity · TUSWill it roast evenly?LAB-MEASURED
Bean-to-bean consistency across the 96 readings. Higher means more uniform, and a uniform lot roasts more predictably. TUS is ranked against other measured lots, so read it as relative.
07Roasting guideHow should I roast it?
Up to four roasting cues from the chemistry, with the data behind each cue. Some lots also show a profile type and each taste as a percentile (0–100) against the reference library. Other lots show rule-based cues marked Preview data.
Where to start
Roasters · buying to roast
Producers & buyers · selling or trading a lot
Where the numbers come from
Every lot is measured the same way.
One green sample per lot, scanned 3 times with near-infrared light. 32 readings per scan, 96 in total.
- Green sampleOne per lot
- 3 scans × 32 readings96 readings, near-infrared light
- 16 componentsThe bars on the card
- 5 tastesPredicted from the components
Uniformity · TUS
How evenly the readings agree, bean to bean. Ranked against other measured lots.
Lot LG-0191%LAB-MEASURED
How the scan works
A single scan captures about 156,000 raw signals, revealing dimensions of the bean that no palate or cupping form can measure directly. The hard part isn't reading them; it's knowing which few actually predict the cup. That's what the uncrack engine does: it finds meaningful patterns and translates them into chemical components, taste axes, and quality signals you can act on.
Try it on a real lot