Cool one room: diagnose the problem, then buy the fix

Almost every cooling guide starts with a product category and works backwards to your situation. That is the wrong direction, and it is why so many people end up owning a machine that does not fix their particular problem.

A hot upstairs bedroom and a hot garage are not the same problem with different square footage. They have different causes, and the correct purchase in one is a waste of money in the other.

Nearly every hot room is some combination of these. Working out which one dominates is ten minutes of work and it decides everything after.

1. Sun through glass. The dominant cause in most rooms with a west or south-facing window. Sunlight deposits energy into floors, walls, furniture and mattresses, and that mass releases it slowly for hours. A room that is unbearable at 10pm was heated at 3pm.
2. Radiant heat from above. An attic in summer sun gets far hotter than the outdoor air and radiates downward through your ceilings all evening. No fan inside the room addresses this — you are being warmed by a hot surface, not hot air. Touch your ceiling at 4pm; if it is warm, this is your problem.
3. Heat generated inside the room. Every watt of electricity consumed in a room becomes heat in that room — exactly, not approximately. A gaming PC under load is 300–500 W of heater. Halogen lighting, a small fridge, networking gear all add up, and in a small or windowless room they dominate.
4. Air that does not arrive or does not leave. Long leaky duct runs to an upstairs room, a thermostat on a different floor, or an interior room with a supply vent and no return so air comes in and cannot escape.

Each has a different fix. Shading beats sun; ventilation and insulation beat attic heat; unplugging things beats internal load; duct sealing or a dedicated unit beats poor delivery. Buying a bigger air conditioner addresses none of them well.

Ordered by effect per dollar rather than by how satisfying it is to buy something. A surprising number of rooms never get past step three.

The first two steps are free and they are the most effective. That is not a rhetorical flourish — preventing heat from arriving beats removing it afterwards every time, because the second requires energy and the first does not.

Step three is the only one that lowers a room’s temperature without a compressor, and it has a hard precondition that catches almost everybody.

The outdoor air has to actually be cooler than the indoor air. Not feel cooler — be cooler, on a thermometer.

Evening air feels cool long before it is cool. The light changes, there is a breeze, and it is still 86°F outside while your bedroom is 82°F. Opening up at that point pumps heat into the house, and it is the most common mistake in cooling without air conditioning. In many climates the crossover comes at 9 or 10pm rather than at dusk.

Once it has crossed: exhaust from the highest, hottest window, with another window open low on the opposite side of the house. Hot air collects high, so exhausting there pulls cooler air in low and drags it through the building. The full method is here.

• Bedrooms are usually a sun problem plus a noise problem. Solvable for most of the season with a blind and a quiet fan; the compressor is step five.
• Upstairs rooms are usually an attic problem, which is radiant and therefore fan-proof. Diagnose before spending anything.
• Apartments are decided by the lease and the circuit before they are decided by cooling. Most lease clauses ban brackets rather than cooling, which rules out less than people assume.
• Garages defeat refrigeration by construction — uninsulated, with a wall-sized uninsulated door. Insulate the door, then move air.
• Dorm rooms are decided by housing policy and a shared 15-amp circuit. The upside is that they are small enough for fans to genuinely work.
• Windowless rooms are an entirely different problem: heat has to go somewhere, and without a window you need to find a route or change expectations.

If none of those describes your situation, start with the general no-AC page, which ranks nine common methods by how much each actually does — including two popular ones that barely work at all.