How we pick

1 — Fix the constraint

"The best bookshelf speaker" has no answer. "Bookshelf speakers under $500" does, because the constraint eliminates candidates. Every article states its constraint in the title and sticks to it.

2 — Build the matrix first

Every candidate's published specification goes into one table before a word of prose is written. Rows that cannot be filled for every contender stay visibly empty rather than being dropped — an absent figure is information.

3 — Use them, and separate what was seen from what was read

Hands-on rows carry the HANDS-ON tag and are written as observations. "Runs noticeably quieter" is a claim this site can make. "Runs 4.2 dB quieter" is not, because that needs a meter this site does not own.

4 — Mark the row winners by hand

A cell is only highlighted when someone has decided which direction is better for that row — lighter is not automatically better, and neither is louder. Rows with no declared winner are left unmarked on purpose.

What this method cannot tell you

• Long-term reliability — nothing here has been owned for years.
• Unit-to-unit variation — conclusions come from one unit each.
• Anything instrumented: frequency response, sound pressure, power draw, filtration. Those appear only where a manufacturer publishes them, taggedSPEC.

Money

The pick is decided before any retailer link is added, and commission rates are never compared. See the affiliate disclosure.