How We Test

The short version

We shortlist from the whole category, not from whatever is easiest to link to. We check the claims that matter against something other than the seller's own marketing copy. We rank on the specifications that predict how long a product lasts and how easily it can be kept alive. We publish the reasoning rather than just the result. And we say plainly what we do not know.

Step 1 — Define the category honestly

Before anything is shortlisted we write down what the category is for. “Best nixie tube clock” means a clock using genuine cold-cathode tubes — which immediately excludes a large number of RGB LED products marketed with the word, several of which outsell the real thing. Defining that boundary first is what stops a shortlist drifting toward whatever is most heavily advertised.

Step 2 — Build the shortlist from the market

We start from the full set of products available in the category, not from a sponsored list. Availability is a filter — a product that is excellent but permanently out of stock does not make a guide useful — but it is never a ranking factor, and neither is commission rate.

Step 3 — Verify the claims that carry weight

Listings are marketing copy. Every specification that affects our ranking is checked against the manufacturer's own detail page, a datasheet, or the documented behaviour of the named component. Two rules apply throughout:

• An unstated specification is treated as the cheap option. A tool chest that does not name its steel gauge is thin steel. A clock that does not name its real-time clock module has the drifting one.
• Anything we cannot verify is reported as unverified, in the entry, rather than repeated as fact.

Step 4 — Rank on what predicts longevity

Every category gets a written checklist before assessment begins, so a product cannot be re-weighted into first place afterwards. The recurring themes:

• Serviceability — can the consumable parts be replaced without specialist tools? Are spares actually available?
• Longevity features — does the design do anything to protect the part that wears out?
• Build and materials — what the enclosure, mountings and tolerances are made of.
• Safety and compliance — relevant certification, and how hazards are enclosed.
• Support and documentation — a manual, a warranty you can claim on, and a company still answering email.
• Value for the category — what you get relative to the realistic alternatives.

Price is deliberately not the heaviest factor. A cheap product that fails in two years is not good value, and weighting price heavily produces rankings that read well and disappoint in practice.

Step 5 — Write it so a sceptic can check it

Every ranked entry states why it made the list, what to check before buying, the specifications we verified, and who it suits. If a product wins on one axis and loses on another, the entry says so — a five-product guide where every entry is excellent is not a ranking, it is a catalogue.

Step 6 — Revisit, and log the change

Guides are reviewed when a product is discontinued or revised, when a significant new entrant appears, when a pattern of reliability complaints emerges, or at twelve months, whichever comes first. Changed rankings carry a dated note explaining what changed and why.

What gets a product demoted or dropped

• It turns out not to be what the category requires — an LED display sold as a nixie clock, for instance.
• A safety issue, or a hazard that is not properly enclosed.
• Consumables that cannot be replaced without specialist equipment.
• A pattern of unanswered warranty claims or support requests.
• A quiet mid-life cost reduction that changes the product we assessed.
• Cloud dependency that turns hardware into a paperweight if a service closes.

What we do not do

• We do not accept payment for placement, position, a score, or inclusion in a shortlist.
• We do not send drafts to manufacturers for approval before publication.
• We do not rank by commission rate.
• We do not reproduce prices or star ratings from retailers — we link to the live listing instead.
• We do not describe a product as tested unless somebody has actually used it.