How we research
Updated
Brew Method Guide compares exact product versions using manufacturer documentation, manuals and clearly identified independent sources. We have not physically tested any product. This page explains what that means for how a recommendation is made, and how to tell us when we are wrong.
Our sources, in order
- Manufacturer documents. The maker's US product page, manual and specification sheet for the exact model. Where a maker publishes several versions, we record which one we describe.
- Certifier lists. A certification is cited only from the certifier's own published list, never from a badge on a box or a retailer's description.
- Independent published tests. Findings from named publications, summarised in a short factual line with the publisher's name, link and date. We do not copy long passages and we do not repeat their scores.
- Owner discussions. Forums and reviews by owners help us find questions to check. A complaint becomes a limitation on our pages only when it appears in at least two independent places, and we write it as a reported complaint, not as our own finding.
An answer from an AI tool is never a source. Neither is a retailer listing. Amazon is used only as a place to buy, never as a source of specifications, prices, ratings or customer reviews.
How a figure is labelled
Each fact on our pages is one of five kinds, and where it matters the page says which.
- Manufacturer-stated: published by the maker, in the maker's own unit.
- Published test: reported by a named independent source, with a link and a date.
- Calculated: worked out by us from stated figures, with the formula shown.
- Editorial judgment: our reading of the facts, written as a judgment and not as a measurement.
- Unresolved or conflicting: two sources disagree, or a fact is not published.
How we extract facts
For each exact model we copy the figures we need from the sources into one record: dimensions, capacity, filter type, what is in the box, what you must buy separately, warranty and so on. Each value keeps the maker's own unit and wording and a link to where we read it, with the date. Some products come in several versions that differ by size, colour, material or country. The record states which version a page describes, and the Amazon button goes to a listing for that same version.
When a fact is not published, it stays blank and the page shows “Not listed”. We never enter a guess, and we never show zero for an unknown. Absence from a maker's compatibility list is recorded as unknown, not as compatible, and nothing is carried over from a similar model.
A model's card shows the maker's own US list price, with the maker named and the date we read it. We do not show Amazon prices, because prices change and Amazon's program limits where they can be shown.
When sources disagree
We keep both figures, say which source gives which, and mark the fact as unresolved. A maker's document usually outranks a retailer's text, but a conflict is still shown, not quietly settled. A conflict can lower how firmly we recommend a model, and we say so on the page. When a maker corrects a figure, we update it and log the change.
How we calculate
The maker's unit is stored and shown first. A conversion is worked out when the page is built, shown with “about” or “≈”, and never rounded into a new claim: for example 40 fl oz (≈ 1.18 L).
A maker's “cup” is a convention that differs by brand, so a cup count always appears beside the maker's cup size, as in “8 cups (5 fl oz each)”. Where a maker does not state a cup size, we say that and do not assume one. Our estimates of how many mugs a brewer fills are labelled as approximate.
Our calculators show their formula and their assumptions, accept your own numbers, and refuse input they cannot use instead of guessing. The cost of paper filters, for instance, is the pack price divided by the number of filters, multiplied by your brews per week and 52, assuming one filter per brew and leaving out shipping, tax and electricity.
Editorial judgment
A comparison is more than a table. Choosing what matters for a given use, naming the main limitation of each pick and saying who should skip it are editorial judgments, and we write them as such. They rest on the facts above and are never presented as measurements.
Physical testing status
We do not physically test products. Our pages do not use first-person testing language, such as claims of having measured a product or of how it behaves in daily use, and they do not describe how a product feels to use, because we have not used it. Every review of a model carries this sentence: “Assessment based on manufacturer documentation and independent published sources; we have not physically tested this unit.” Any performance figure that appears on a page is attributed to the maker or to a named published test. If we ever test something ourselves, the page will say exactly what was done and label the result as ours.
How picks are ordered
Picks are ordered by how well they fit the use the page names, such as a small counter or a one-person routine. The deciding criteria are written on the page, next to each pick, along with its main limitation. We publish no numerical scores and no star ratings.
- A higher price is not treated as better fit.
- The commission Amazon may pay does not change which model is chosen or where it sits.
- No maker or retailer pays to be included or ranked higher.
- A model with no usable Amazon listing can still be our pick. The page says that no Amazon route is offered.
- A discontinued or unavailable model gets no buy button.
The affiliate disclosure explains how the commission works.
Software and AI tools
We use software, including AI tools, to help find sources, extract figures and draft text. These tools are never a source. The editor is responsible for what is published, and each figure on a page is tied to the maker document or published source it came from.
How pages are updated
Every buying guide, review and comparison shows the date it was last updated, and every product record carries the date we last read the maker's US page. We recheck a page when a maker changes a product, manual, version or list price, when a reader reports an error, and when we see that a source has changed. If we can no longer confirm a model is on sale, we say so on the page and remove its buy button. Connected pages, tables and calculators are updated together, so that one figure never differs between two places.
Sources and corrections
Every buying guide, review and comparison lists its sources, with links, near the end. To report an error, email contact@brewmethodguide.com with the page address, the claim you think is wrong and, if you have one, a source such as a manual or the maker's page. For each report we record the page, product and claim, the source supplied, our decision, the editor and the date. If the report is right, we correct the page and everything that uses the same figure. We publish every correction in the log below, and a change of recommendation is stated on the page itself.
Corrections log
No corrections have been logged yet. Each entry will show the date, the page, what was wrong, the source that settled it and what changed.