Most coffee “reviews” online are rewritten manufacturer copy with an affiliate link attached. The writer never touched the product. Nothing was tested. The recommendation exists because the commission exists.
We think you should be able to check whether that’s what you’re reading. So this page sets out exactly how we work, what we actually do before recommending something, what we can’t test, and how we make money.
If we ever fail to live up to what’s on this page, we’d like to hear about it. There’s a link at the bottom.
Status
Coffee and bean recommendations: active.
Equipment reviews: not yet published. We will not publish equipment reviews until we are testing equipment to the standard described below. When that begins, this page will be updated and dated.
We’d rather publish nothing on grinders than publish a grinder roundup assembled from spec sheets. If you’re reading this and there are no equipment reviews on the site, that’s why.
The short version
- We buy the coffee we review.
- We control brewing variables so comparisons mean something.
- We taste blind where it’s practical to do so.
- We say when something isn’t worth buying, including when we’d earn a commission on it.
- We tell you what we haven’t tested and what we’re unsure about.
- We correct mistakes publicly and note that we’ve done it.
- We never accept payment for a positive review. Not now, not later, not at any price.
Who does the testing
Testing and reviews are carried out by one person. Every review carries a byline. If you want to know who formed an opinion, it’s on the article.
How we evaluate coffee beans
This is what we do now, and it’s the basis of every bean, roaster and subscription recommendation on the site.
Controlling the variables
Comparing two coffees only tells you something if everything else stays the same. For every coffee we assess:
- Same grinder, cleaned between coffees, with grind size adjusted only to compensate for bean density
- Same water, filtered, with consistent mineral content. Water chemistry changes coffee more than most people realize, so we don’t let it vary
- Same ratio unless we’re deliberately testing ratio
- Same brew temperature, measured rather than assumed
- Same vessel and same technique
Where a coffee is being assessed for a specific method, we brew it that way. Where we’re assessing it generally, we use pour-over as the baseline, because paper-filtered brewing shows the most detail.
The testing window
We don’t judge a coffee from a single cup on a single day.
Coffee changes as it ages. A bag can be tight and closed at five days off roast, at its best around two weeks, and fading by four. Reviewing on day one tells you very little.
We brew each coffee multiple times across the life of the bag, typically an early session, a mid-window session and a late one, and we note how it develops. A coffee that’s brilliant for four days and dull thereafter is a different proposition from one that holds up, and you deserve to know which you’re buying.
Blind tasting
Where practical, we taste blind, with coffees prepared and labeled by number and the identity revealed afterward.
Brand names, prices and pretty packaging genuinely influence perception. Blind tasting doesn’t eliminate that, but it removes the most obvious source of it.
We’re straightforward about the limits. With a single taster, “blind” means the identity is concealed at the point of tasting, not that we’ve run a controlled study. It’s a discipline, not a laboratory.
What we’re actually judging
- Aroma, both dry grounds and wet
- Flavor, what’s there and whether it matches what the roaster claims
- Acidity, quality and structure rather than just intensity
- Body and texture
- Sweetness
- Balance, meaning whether the parts work together
- Finish, what’s left and for how long
- Consistency, whether it performs the same way twice
- Value, quality relative to price rather than quality in isolation
We also record the practical things: roast date on arrival, packaging quality, whether stated information was accurate, and how quickly it shipped.
What we can’t do
Being honest about the limits is part of the method.
Taste is subjective. We can tell you a coffee is well-made, fresh and true to its description. We can’t tell you that you’ll like it. Where we suspect a coffee will divide people, we say so.
We can’t taste every lot. Coffee is agricultural. The Ethiopian we assessed in March is not the identical coffee that roaster sells in October, because it’s a different harvest and sometimes a different producer. Our assessment describes what we tasted, not a permanent guarantee.
We’re not a cupping lab. We don’t assign SCA cupping scores, because we’re not certified to do so and pretending otherwise would be dishonest.
One palate has limits. We’re one operation, not a panel.
How we will test equipment
Not yet active. This is the standard we’re committing to before any equipment review appears.
How we choose what to test
By what readers are actually searching for and asking about, not by what has the highest commission rate. We use Search Console data and reader questions to decide what to buy.
Testing period
Minimum 30 days of regular use before we publish anything. Some categories need longer.
- Grinders: at least 5kg (11lb) of coffee through them, to see how they behave once seasoned and whether retention or consistency degrades
- Espresso machines: a minimum of 60 days, because temperature stability and reliability problems rarely appear in week one
- Kettles, scales, brewers: 30 days minimum, longer where durability is the main question
A product used twice is not a product tested.
Controls
Same principle as coffee: one variable at a time. When comparing grinders, everything else stays fixed, including beans, water, ratio, brewer and technique. Where a measurement is possible, we measure it rather than describing an impression.
What we measure versus what we judge
Measured: grind consistency and retention, dose accuracy, time to temperature, temperature stability, brew temperature at the bed, weight, footprint, noise.
Judged: build quality, usability, how annoying it is to clean, whether the design fights you, whether it’s still pleasant to use on day 30.
Both matter. We’ll be clear about which is which, because a measurement is a fact and a judgment is an opinion.
Long-term follow-up
Where a product stays in regular use, we revisit it. A machine that fails at month eight is more useful information than a machine that impressed us in week one, and it’s the kind of thing most reviews never report, because the reviewer moved on.
What we can’t do
We can’t test everything. We’re one operation with a finite budget. A roundup covers what we’ve tested. If a well-regarded product isn’t in it, that usually means we haven’t bought it, and we’ll say so rather than including it on reputation.
We can’t predict reliability from a sample of one. If our unit fails, we report it as our unit failing, not as a verdict on the product line. We’ll say when we’re describing a single data point.
We can’t replicate every setup. Your water, your beans, your kitchen and your technique differ from ours.
Where our products come from
We buy most of what we review. Where a manufacturer or roaster provides a product for review, we say so clearly at the top of that article. A sample buys coverage consideration and nothing else: no editorial input, no approval rights, no guarantee of a positive verdict, and no obligation to publish at all.”
Whichever you choose, these hold:
- No manufacturer sees a review before publication.
- No manufacturer has approval rights over anything we write.
- We don’t return samples in exchange for favorable coverage, and we don’t accept conditions attached to a sample.
- If a company withdraws cooperation because they didn’t like a review, we’ll report that too.
How we make money
Directly, because you should be able to weigh it.
Affiliate commissions. Some links on this site are affiliate links. If you buy through one, we may earn a small commission at no additional cost to you. Affiliate links are marked and disclosed at the top of any article containing them.
Display advertising. We run ads. Advertisers have no influence over editorial content and no advance sight of it.
What this doesn’t mean:
- We don’t rank products by commission rate.
- We don’t take payment for a review, a placement, or a position in a roundup.
- We publish negative conclusions about products we could earn money from, and “don’t buy this” is a legitimate outcome of a review.
- We recommend not upgrading when not upgrading is the right answer, even though that earns us nothing.
A site that only ever recommends things isn’t reviewing anything.
Full details are in our affiliate disclosure.
Scores and ratings
We don’t use numeric scores.
A grinder rated 8.4 out of 10 implies a precision that doesn’t exist. It suggests the reviewer measured something they didn’t measure, and it collapses genuinely different questions, including build quality, value, grind consistency and usability, into one number that answers none of them.
Instead we tell you what a product is good at, what it’s bad at, who it suits, and what we’d buy instead at the same price. That takes longer to read and it’s more useful.
Where we do name a best pick, we say what it’s best for.
Keeping reviews current
Coffee gear changes and coffee itself changes every harvest.
- Every article shows a published date and a last-updated date.
- When we substantially revise a review, we note what changed and why.
- When a product is discontinued, we mark it rather than quietly leaving it in a roundup.
- We don’t review discontinued equipment as though it’s current, and we don’t leave old recommendations standing when better options arrive.
Corrections
We get things wrong. When we do:
- We fix it.
- We note that we’ve fixed it, and what was wrong.
- We don’t silently edit an error out of existence.
If you’ve spotted something wrong, whether a factual error, an out-of-date claim, or a recommendation that no longer makes sense, please tell us. We’d rather hear it from you than not know.
Hold us to this
If anything on this site doesn’t match what’s on this page, that’s a problem we want to know about.