How CoffeeLabAdvisor knows what it knows
CoffeeLabAdvisor does not try to evaluate every machine, grinder, roaster and drink in coffee. It does try to recognise the ecosystem it writes about — and to be exact about which of four things is true of any one thing in it. Every Advisor answer, every card on a review, every specification record carries one of these four labels.
The four labels
Independently evaluated
We evaluated it under the published methodology: a review, a profile with a documented specification, a brand hub, a grinder pairing page, or a CoffeeLab Award. The Advisor sends you to that page and adds what it is best suited for and when you might consider another option — sentences taken from the page itself. Scores, where they exist, appear on award pages only.
Specification verified
We read the maker's, or a listing's, specification and dated the reading. The record says what the thing is and who it may suit — each line derived by a written rule from a specification field (the line is marked as coming from the specification) — and it says plainly that we have not evaluated it. A specification record never says verdict, never says we tested, never lists pros and cons, and never describes performance. After a year the reading is marked due for a re-check.
Identified
We can identify it and say where it appears on our site: a verified listing on a guide, a roaster's bag on our roster, a machine named on a city award page. We do not say who it suits, because we have no basis to.
Not yet evaluated
We have not evaluated it and it is not something we can identify. The Advisor says so, shows the nearest things we do cover, and logs the words you typed — no name, address or device details — as a signal of what to write next.
How the who-it-may-suit lines are derived
Fit lines on specification records are not written per product and are not generated by software at request time. Each comes from a rule that is written down with the field it reads, the sentence it produces, the reason, a confidence level and the name of the person who reviewed it. A rule with low confidence renders as may suit. Rules stay silent when a field is missing. Before each expansion, every rule's output is read against the machines and grinders we have evaluated; a rule that would mis-describe one of them is downgraded or retired.
What the Advisor is not
- Not a ranking. It compares nothing and scores nothing.
- Not a review of anything it has not evaluated. The label comes first on every answer so the two cannot be confused.
- Not a catalogue of everything. Breadth of recognition, depth of evaluation where warranted — and an honest not yet in between.
Coverage versus evaluation
CoffeeLabAdvisor covers the specialty coffee value chain — equipment, coffee, drinks, venues and the brands behind them — in the sense that it can identify and contextualise the things in it. It evaluates a documented subset of that chain under one standard. The Advisor is where the difference is made visible.
The specification records so far
Machines we hold a dated specification for but have not evaluated — each one a record, not a review:
- Breville Dual Boiler — specification read on wholelattelove.com, 2026-08-18
- Lelit MaraX V2 — specification read on wholelattelove.com, 2026-08-18
- Rancilio Silvia Pro X — specification read on ranciliogroupna.com, 2026-08-18
- Lelit Bianca V3 — specification read on wholelattelove.com, 2026-08-18
- La Marzocco Linea Mini — specification read on home.lamarzoccousa.com, 2026-08-18
Sending documentation
Makers and shops can send specification sheets or documentation to support@coffeelabadvisor.com. A verified document refreshes a record's reading date and is credited on it. This is free, optional, and has no bearing on any award.