How we score
The Bench Score methodology
Every product we review gets a Bench Score out of 10. The score is our editorial judgment, made by us, and informed by two things: the product's documented specifications, and the consensus of owner feedback we research from public forums, owner communities and video reviews. It is not an average of customer star ratings, and we do not accept payment to change it.
We score five dimensions from 0 to 10, then combine them with the fixed weights below. Showing the parts, not just the total, is the point: you can see exactly why a machine landed where it did, and weight the dimensions differently if your priorities differ from ours.
The five dimensions
| Dimension | Weight | What it measures |
|---|---|---|
| In the cup | 25% | How good the coffee can actually get. |
| Value | 25% | What you get for the money. |
| Ease of use | 20% | How forgiving it is day to day. |
| Build & longevity | 20% | How well it is made and how long it lasts. |
| Versatility | 10% | Range of drinks and techniques it supports. |
Principles
- +No score is reverse-engineered to flatter a product or its price. We score the dimensions first; the total is whatever the maths gives.
- +Scores are judgments, not lab measurements. Where we rely on owner consensus rather than hands-on testing, we say so on the page.
- +Scores change. As products get updated, prices move, or owner feedback shifts, we revise them, and the date on each review reflects the last review.
- +New releases get a "first look" score, marked provisional and based only on the published specifications. We revise it once real owner feedback exists, so an early score is a starting read, not a final verdict.
- +Affiliate commissions never influence a score. They fund the site; they do not buy a number.
Named data points
We name our numbers, and every name is a plain-English claim the number literally supports. If you quote one, this is what it means.
- Bench Score
- Our editorial score out of 10: five weighted sub-scores set by us, never aggregated from users.
- Provisional Bench Score
- A Bench Score built from published specifications alone, on a first look, before owner consensus is researched.
- Owner praise and Known gripes
- Recurring positive and negative themes from public owner discussion, paraphrased, with the sources and the check date shown on the page.
- Buy-at price
- The price we would pay for a machine, published months before Black Friday so a discount can be judged against it.
- Cost per descale
- A verified retail price divided by the maker's stated doses per pack.
- Espresso Water Score
- A city's tap water scored 0 to 100 against the SCA brewing standard from its utility's own published figures.
- Descaling interval
- Weeks between descales implied by a city's carbonate hardness for a machine pulling two double shots a day: a comparison figure, not a prescription.
- The ratchet
- The finding that US retail coffee follows commodity rises faster and further than commodity falls.
- Creep rate
- Days earlier per year that a chain's pumpkin lineup launched, from a straight-line fit with a 95% interval.
Names are defined once, in a register, before they are used in data or copy. A name that would imply a measurement we do not make is not used.