Cold brew is a low-efficiency extraction. Without heat, water pulls out far less of what is in the bean, and specifically pulls out much less of the acidic, aromatic, fruity material that makes a light roast exciting. What survives is the heavier stuff: chocolate, caramel, nut and body. So the bean selection logic is simple and slightly counterintuitive — the coffees that shine in a pour-over are usually the wrong choice here.
Buy medium to dark roasts with chocolate, caramel or nutty character. Buy coffees with body. Skip the bright, floral, light-roast single origins, not because they are worse but because cold water leaves the very qualities you paid for sitting in the grounds. The picks below are ranked on exactly that, plus value per bag — because cold brew uses roughly four times the coffee per finished cup of a drip machine, so cost per pound matters more here than anywhere else on this site.
If you have not chosen a brewer yet, our cold brew maker ranking covers the immersion systems and the ratios that go with them.
Why cold brew changes which beans work
Solubility is temperature-dependent. At brewing temperature, hot water dissolves a wide range of compounds including the delicate aromatics and bright organic acids that give a light-roast Ethiopian its jasmine-and-lemon character. At room temperature or in a fridge, a large share of those compounds simply never make it into the water — which is exactly why cold brew tastes smoother and less acidic.
The corollary is the useful part: if you cold brew a delicate light roast, you are paying for qualities that stay in the grounds. What does come through is the heavier, less temperature-sensitive material — the chocolate, caramel, nut and body of medium and darker roasts. That is not a compromise; it is choosing the right instrument.
Ratio, grind and time
Grind coarse— French press coarse or a shade coarser. Fine grounds do not make cold brew stronger, they make it muddy and hard to filter, and they over-extract at the edges while the middle is still going. This is the one grind setting most home grinders find hardest to produce evenly, which is why a purpose-ground bag is a legitimate shortcut.
Ratio: start at 1:5 by weight for a concentrate you dilute, or 1:12 for ready-to-drink. Time:12–18 hours at room temperature, 18–24 in the fridge, since colder water extracts more slowly. Then filter and stop — coffee left on the grounds past your target keeps extracting and turns woody.
Budget for the volume
This is the practical reason big bags appear on this page. A drip machine uses roughly 60g of coffee per liter of finished coffee; cold brew concentrate at 1:5 uses around 200g per liter of concentrate. Even after dilution you are typically using two to three times the coffee per finished cup. A $25 specialty bag that lasts three weeks in a drip machine lasts about one batch here. Factor that in before you commit to an expensive bean.
Roast date matters, but less than usual
Cold brew is the most forgiving method on this site for slightly older coffee. A long, gentle extraction does not depend on the CO2 behavior that makes stale beans pull badly under espresso pressure, and the compounds that fade first are largely ones cold water was not going to extract anyway. Fresher is still better — but a three-week-old bag makes perfectly good cold brew, and that is why a pre-ground purpose-made option earns its place here when it would not on any other page.
What to avoid
Light, floral single origins — you will not taste what you paid for. Very oily, aggressively charred beans— the ashiness concentrates unpleasantly over 16 hours. Pre-ground drip coffee— it is ground far too fine, which clogs every filter and produces a silty, bitter brew. And flavored coffees tend to taste synthetic when concentrated. If you want your cold brew sweet or vanilla-flavored, add it afterwards.
How we picked
We did not lab-test this gear
Everyone in this category says they tested twenty machines. We have not lab-tested any of these, and we say so. What we did instead: compiled the published manufacturer specifications, read the manuals, ran the math where there was math to run (heat-up time, pressure, dose capacity, grind range, cost per cup), and scored each pick against a published rubric. The scores are judgments from documented research — not measurements we took, because we do not have a lab and we will not pretend we do. Where a number came from someone else's work, we name them in Sources.
