How we rank

What we rank on

Two things decide whether you keep wearing a pair of workout earbuds: whether they stay in, and whether they survive sweat. Together those are 45% of every score on this site, on purpose. A pair that sounds wonderful and relies on canal friction ranks below a plainer pair with a hook; a pair with a splash rating ranks below a pair rated for water jets. After that come the things the buyer this site serves cares about next: awareness of the room or road (for open designs) or isolation (for sealed ones), sound, calls in wind, battery, and value.

What the published score means

Every product on this site is one we would tell a friend to buy. Weak picks and anything we cannot find in stock get replaced, not listed with a low mark — so the published score tells you how strong a recommendation this is, not how a product compares with hardware we would never put in front of you. The three bands:

Products that would fall below 8.4 are not recommended on this site. They are replaced with a stronger in-stock alternative rather than published with a low number, and the build fails if one slips through.

One site-wide scale does the conversion. The rubric below produces a raw weighted mean on its own 0–10 criterion scale; the published score is that figure mapped linearly onto 8.4–9.8 across the whole catalog at once. The mapping is a single formula applied to every product, it is strictly increasing, and it preserves the distance between any two products — so the order of every ranked list is identical before and after, no product is ever nudged on its own, and a product shows the same score on every page it appears on. The criterion marks in the “how the score is built” panel on each card are the raw rubric marks, which is why they read lower than the published total.

The rubrics and their weights

Each product is scored 0 to 10 on each metric and the raw overall is the weighted mean, computed by the same code that renders the card, so a page can never show a total that disagrees with its parts. Open-ear and sealed designs get different third metrics, because they do different jobs:

What each score is built from

• Fit security is judged by the retention system, not by feel: a neckband scores 10, a sealed ear-hook 9 to 10, an open ear-hook 9, a clip-on 8, a wingtip 8. Adjustable hooks and multiple cushion sizes don't raise the ceiling but do make the score reachable on more ears. A plain in-ear tip with no hook or wing is not listed on this site.
• Sweat and water resistance is the maker's IP rating, read literally from its page: IPX4 or IP54 scores 5 to 6, IP55 scores 8, IP66 scores 9, immersion ratings (IP57, IPX7, IP67, IP68, IPX8) score 10. A rated case adds nothing to the score but is noted on the card. Where a maker publishes two ratings on one page, we score the lower.
• Awareness (open designs) credits a completely unobstructed canal (neckbands, 10) slightly above a speaker sitting in front of it (hooks and clips, 9). Noise control and transparency (sealed designs) credits ANC plus a transparency mode (9) above transparency alone (6).
• Sound and leakage credits dual or large drivers and high-bitrate codecs, docks single-driver budget designs, and scores bone conduction at 4 to 6 because the category has little bass; SoundGuys' published finding on that is cited on every relevant page. The sound score reads the driver system and the measurements published by the reviewers we cite.
• Calls and wind credits a published wind rating (Shokz 25 km/h, Suunto 30 km/h), a bone-conduction voice mic, or a boom, above an unrated array; mic count matters less than a stated wind strategy.
• Battery and charging is the per-charge figure, then quick charge, then the port (USB-C over a proprietary cable).
• Value is our reading of the maker's list price against the spec, on the day checked. Prices themselves are never printed.

Where the figures come from

The maker's own product page, read on the date shown on each card, and linked from every value. Shokz publishes a single comparison table for its lineup; that table is the source for every Shokz figure on the site. Where a maker's page can't be read without scripts (Nothing) or blocks automated fetches (Sony), we cite a named independent review and say so on the card. Where a figure isn't published, the table shows a dash. The full fetch log is kept in the site's repository. Standards (IEC 60529) and health sources (NIDCD, WHO) are cited on the guides that use them.

What a score is built on

Every pick is researched from the maker's published specifications and manual, its sweat and water rating is read against the IEC 60529 standard, and independent findings are quoted with the name of whoever published them (“SoundGuys measured 7 hours 10 minutes”). Health and hearing facts are cited to NIDCD and WHO. Every pick also carries a “good to know” line naming its one real trade-off, so you can see which pair fits your use case before you spend anything.

How a product gets removed

When the maker's page disappears or the product is discontinued, it comes off the site rather than staying with a dead link. Flagship roundups are rechecked every 60 days, other roundups and comparisons every 90, guides every 180 or when a standard changes. Corrections are logged on the page they touch.

Commissions and rankings

Openear Club earns affiliate commissions, mostly as an Amazon Associate. Commissions never set a score or decide a pick, and no brand pays for placement. Every outbound money link is marked and disclosed. Details in the affiliate disclosure and the editorial policy.