How we rank

What we rank on

Three things decide whether a cooler or a tumbler gets bought once: how long it actually holds ice or cold, what the maker has put on paper about fixing it when it fails, and whether the parts that wear out can be ordered. Capacity per dollar, build quality and a plain editorial call on whether the thing suits a buy-it-for-life owner round it out. Those are the axes the site's brief named, and the axes a person who hates rebuying actually shops on. Nobody buys a second cooler because the first one had the wrong logo.

The scale, and what a score means

Scores on this site run from 8.4 to 10, and that is deliberate. Nothing is listed here unless it is worth buying: a product that would land below 8.4 gets replaced by a stronger one rather than published with a weak number. So the scale measures how strongly we recommend something among things already worth recommending, not how it compares to every cooler ever made. Read it like this:

• 9.5 to 9.8 — category-defining. The measured ice hours, the warranty term and the parts catalogue all line up. If you buy one of these you are very unlikely to buy another.
• 9.0 to 9.4 — outstanding. Strong on the axes that matter, with one trade-off you can see on the card (usually price, size or a shorter term).
• 8.4 to 8.9 — excellent, and strongly recommended. The best thing in its class for a particular job, where the class itself has limits — a twelve-can soft bag cannot hold ice like a rotomolded chest, and we do not pretend otherwise.

Below 8.4 is not a bad score; it is not on the site. If a product cannot earn 8.4 on the rubric it comes off the list and a stronger, in-stock alternative takes its place.

The scale was recalibrated site-wide on 4 November 2026. The rubric criteria and weights below did not change and no judgment about any product changed: a single monotonic conversion was applied to every score at once, so every ranking and every gap between two products is exactly what it was. The old axis compared a lunch bag against a 48-quart chest on the same absolute ice line and published class-leading gear at 5 or 6 out of 10, which read as “mediocre” for products we would recommend without hesitation. The conversion is published too, for anyone who wants to check it: a rubric mean of 3.4 (the lowest on the site before the change) maps to 8.4, a perfect rubric 10 maps to 10.0, and everything in between falls on the straight line between them.

The rubrics and their weights

Each product is judged on each metric and the overall score is the weighted mean, computed by the same code that renders the card, so a page can never show an overall that disagrees with its parts — expand “How the score breaks down” on any pick and the numbers add up. Coolers and drinkware use the same weights so a 9.2 means the same thing in both hubs; ice packs and parts use a shorter rubric.

What each score is built from

The per-metric numbers below are on the same published scale as the overall. The 8.4 floor applies to a product's overall score; a single line inside the breakdown can sit lower than that — a one-year warranty is a one-year warranty, and the card shows it.

• Ice retention (reported), 30%. Hours or days to 40°F and 50°F as measured by a named third-party test, cited on the card. A measured figure beats a maker's claim in either direction. Where no test exists, the product is scored on its published insulation thickness and construction, and the card says which. Maker claims (“up to 10 days”) are shown, labeled, and never scored as measurements.
• Warranty term, 20%. Years, from the manufacturer's published warranty page, quoted and linked with the date we read it. Lifetime scores 10; ten years 9.8; five years 9.3; three years 8.8; one year 8.3; 90 days 7.8; a term we could not read on a fetchable page scores as unknown (8.1–8.3), and the card says “unverified.”
• Repairability, 15%. Are the wear parts sold? Latches, gaskets, drain plugs, handles and feet for coolers; lids, gaskets and straws for drinkware. First-party parts on the maker's site score 9.8–10; parts visible only through a warranty claim or third-party listings score 8.8–9.3; nothing listed scores 8.1–8.5.
• Capacity per dollar, 15%. Rated quarts (or ounces) divided by the maker's list price at time of writing. The list price is used only for the ratio; the price you see on a pick is the live Amazon price with the date it was last checked.
• Build & carry, 10%. Rotomolded vs injection-molded, latch and handle design, empty weight, wheels, lid type, from published specs.
• Fit for the buy-once buyer, 10%. The editorial call: does this get bought once? A short warranty, a sealed lid or an unrepairable zipper costs points here even when the ice hours are good.

The tests we cite, and why

Every ice-retention number on this site comes from a publisher that measured it, credited by name. These are the publishers we cite, what they measured, and where:

• OutdoorGearLab: iButton sensors in a controlled room, hours to 40°F and 50°F, 23 hard coolers and 19 soft coolers in its 2026 tables. Source.
• Outdoor Life: one-third-capacity ice load at 60–70°F ambient, hours to melt-out, May 2026; ice packs by infrared thermometer after a 36-hour freeze. Source.
• Treeline Review: two years of daily tumbler use, 180–200°F hot fills, ice tests, dishwasher tests, August 2026. Source.
• Outside Online (Outside Lab): a 96-hour dark, unopened lab test with block ice plus a 3-day backyard test, Yeti Tundra Haul vs RTIC 52, 2026. Source.
• Gear Patrol: 72-hour patio test in ~90°F heat with ThermoWorks sensors every 4 hours, Tundra 45 vs RTIC 45. Source.
• Switchback Travel: qualitative field use; the source of the honest 'expect about half the claimed retention' framing. Source.

When tests disagree (Gear Patrol found the Tundra 45 and RTIC 45 within 2.4°F at 72 hours; Outside Lab found the RTIC fully melted at 96 hours while the Tundra Haul still held ice), we show both and explain the protocol difference. See how ice-retention tests work.

How a warranty term gets onto the page

We read the manufacturer's own warranty page, quote the term as written, link it, and record the date. Where the maker blocks automated fetches, we say what we used instead: YETI's terms are quoted from a copy of YETI's warranty page republished by an authorized dealer and cross-checked against Outdoor Life's 2026 reporting; Hydro Flask and Coleman could not be read at all and are marked unverified. Terms change (RTIC's hard-cooler term has moved more than once), so every warranty page on the site carries the date it was read, and readers who spot a change can tell us.

What every figure on a page rests on

• Every measurement is attributed to whoever made it, by name and link, so you can read the test yourself.
• Every spec, warranty term and ice hour comes from a source we read and dated. Where a source has no figure, the page says “unverified” rather than filling the gap.
• Prices are live and dated: each pick shows the Amazon price with the date it was last checked, and the button goes straight to the listing.
• The visible score is editorial content, published with its weights so you can see how a pick was reached.

How a product gets removed

A product leaves a ranking when the maker stops listing it, when it can't be sourced to a maker or dealer page, when the Amazon search that its button points to stops returning it, or when a reader shows us a figure we printed is wrong and we can't re-verify it. Changes are logged on the page.

Commissions and rankings

The site earns Amazon Associates commissions on qualifying purchases. The rubric is fixed before a product is scored, the scores are published with their weights, and a product's commission rate is not an input to any of them. See the editorial policy and affiliate disclosure.