How we research
Updated October 21, 2026
We don't test welders. We read everything published about them, check it against itself, and show our work.
The method, for every guide
Set the criteria first
Before any research comes back, we write down who the guide is for and how machines will be ranked, so the picks can't be fitted to the data afterwards.
Build a wide candidate list
Usually 10 to 15 machines: what ranks in search, what Amazon US sells, what makers currently offer, and what owners recommend.
Read the primary sources
Every spec comes from the maker's spec sheet, manual or product page, with the date we read it. When a maker's documents disagree, the manual wins and we say so. When a maker's site can't be reached, we read a copy of the manual hosted elsewhere and the sources list says so.
Read what owners report
Forums, Reddit and owner videos, read for problems and praise that recur. Quotes are short, word for word, and linked.
Check every number twice
Before a page goes up, a separate check compares each figure and claim on it with its source, and each pick's store listing is confirmed from the US.
What we don't do
- Test machines ourselves, or imply that we did
- Invent reviewers, experts or authors
- Show prices; they change too often to be honest
- Repeat Amazon star ratings
- Let an affiliate program decide a ranking
- Hide a maker's conflicting numbers
How we make money
Some links earn us a commission when you buy, at no extra cost to you. Whether a brand runs an affiliate program never decides where its machine ranks: in our first guide, three of the four brands with affiliate programs didn't make the picks. As an Amazon Associate we earn from qualifying purchases. The full list of programs is on the disclosure page.
Corrections
If a number on one of our pages is wrong, tell us and send the source: contact@bestweldingequipment.com. We check it, correct the page and update its date.