Roaster, origin, producer, process, variety, price, freshness, evidence, community availability, and brew outcomes connected clearly.
BEAN ON BAR INTELLIGENCE
Helpful coffee guidance
without pretending.
The app’s intelligence layer is deliberately explainable: what the bag says, why a Context Score moved, why a recipe was chosen, what changed after tasting, and what still needs a human correction.
Filter and espresso recipes start from visible bean clues, then tune temperature, grind range, ratio, timing, and next-brew adjustments.
Saved recipes, dose changes, grinder notes, ratings, and taste feedback attach to the bean so your second brew starts smarter.
Context scoring shows visible label signals, while cited Taste Signals remain separate from the score.
SCORING METHODOLOGY
Two signals,
not one verdict.
The methodology separates what a listing discloses from what others have said about taste. Context Score measures visible information. Taste Signals show cited recognition, reviews, or roaster-stated cues without inflating the score.
EXAMPLE CONTEXT SCORE
Kenya Kiambu AA
85Clearly documented
This is not a review claim. It says the listing gives strong context: origin, variety, process, freshness, seller, price, weight, and tasting notes are visible. A separate badge can still say “Strong tasting lead” when cited evidence supports it.
Provenance
Country, region, producer/farm/station/co-op, and variety make the bean easier to understand and compare.
Process & roast
Process, roast profile, and known roaster or seller set fair expectations before tasting.
Freshness context
Printed roast dates score highest. Current cafe or roaster listings can receive a clearly labelled estimated freshness signal.
Buying context
Price, weight, and tasting-note specificity help drinkers compare beans without treating price as quality.
Taste Signals
Cup of Excellence, Coffee Review, user-cited evidence, and roaster-stated scores appear as badges, not score boosts.
Suppression rules
If origin and process are missing, seller context is missing, or source confidence is low, the number is hidden rather than over-claiming.
It is about how much we can see, not how good the coffee is.
Bean On Bar reads what a bag label or public listing discloses — origin, process, variety, roast date, price, weight, and tasting notes — and scores how complete that picture is. A high Context Score means a well-documented coffee. A low or hidden score means we could not see enough to score it fairly.
Evidence stays visible, but separate.
When a credible source has assessed a coffee — a competition result, published review, linked user citation, or roaster-stated cupping score — Bean On Bar shows it as an attributed badge. We do not fold it into the Context Score, and we do not make one up when nobody has cited one.
The method is cautious by design.
Co-ops and washing stations receive full provenance credit. Blends need their own component-disclosure rubric. Decaf method is helpful context, not a requirement. When the same bean appears through multiple sellers, the goal is one complete record with all source links, not duplicate entries.
BEAN INTELLIGENCE DIRECTORY
From scanned bag to
clearer decisions.
These prototype records reuse the community bean dataset. As people scan, brew, report, and correct beans, the directory can become a clearer guide to what is well documented and worth exploring.
PRICE BENCHMARKS
Make value visible,
not mysterious.
Price context should help drinkers compare bag sizes and currencies without pretending every expensive coffee is automatically better or every cheaper coffee is a bargain.
Prototype estimates use sample prices for demonstration. In the app, price-per-gram comparisons should stay grounded in real labels and the user’s own saved history.
LEARNING SIGNALS
Every cup can improve
the next decision.
The most useful signals are practical: what was scanned, what was corrected, what recipe was chosen, how the cup tasted, and whether the coffee was worth brewing again.
Drinkers scan and correct
Editable OCR keeps the user in control and teaches the app which label patterns matter.
Brews become evidence
Ratings, taste feedback, recipe choices, and grind notes make the next brew more useful than a generic recipe search.
Discovery stays source-linked
Community listings point back to cafes and roasters so availability can be checked instead of guessed.
Corrections protect trust
Users, cafes, and roasters can flag stale availability, unclear data, or wrongly attributed listings.
COMMUNITY HEALTH
Signals that show
people are helped.
As the community grows, the healthiest signals are simple: people identify beans more confidently, brew better first cups, correct unclear data, and rediscover coffees they loved.
Shows whether users repeatedly use the core wedge.
Shows whether Bean On Bar becomes a coffee memory, not just a one-time scanner.
Shows brew guidance is useful at the moment of making coffee.
Shows the local discovery graph is being built by users.
Shows users are building reliable recipes for specific beans.
Lets drinkers, cafes, and roasters flag outdated availability, unclear recipes, or label details that need fixing.