Where natural cooling actually works (and where it doesn't)

Published 2026-05-14 · by WindowWise

You've heard the advice: open the windows at night, close them before the heat hits. Free cooling.

It's mostly right. But "mostly" is doing a lot of work — there are climates where the strategy saves real money, climates where it's marginal, and climates where it imports more heat and moisture than it removes. We replayed our engine over a year of historical hourly weather for 30 cities to see where the math actually pencils out.

Here's what the data shows.

The three patterns the data surfaces

Pattern 1 — Mountain-diurnal-swing climates: the strongest fit on the planet

Cities at high elevation in mid-latitudes have something most US climates don't: huge swings between daytime highs and overnight lows.

Kathmandu alone produces more annual opportunity-hours than any city in our US sample. Mountain markets are under-recognized for this — most US energy-savings advice defaults to coastal-city assumptions, and misses the strongest diurnal-swing zones entirely.

If you live in the Mountain West, the Andes, the Atlas, the Anatolian or Iranian plateaus, the Himalayan foothills, or Ethiopian highlands: the strategy works for you almost any day there's daylight. Six to nine months of usable opportunity per year.

Pattern 2 — Mediterranean climates: year-round mild

Coastal California, Spain, Italy, Greece, much of the Levant — these climates have mild year-round temperatures with daily swings that consistently cross our default 68-74°F comfort range.

The Mediterranean basin is the "always-on" market for natural cooling — useful 7-10 months a year. Not the strongest single-day savings, but the steadiest.

Pattern 3 — Hot-humid climates: limited and tricky

Houston, Miami, the Gulf coast, tropical international — opening windows in these climates often imports more moisture than the cooling is worth. Air conditioners spend significant energy DE-humidifying air; a window opened on a 75°F-but- 80% humidity night undoes that work.

Our engine doesn't yet gate on humidity (it's an open roadmap item), but the data suggests these climates need much narrower open-windows criteria than the dry-climate equivalent. Houston / Miami numbers in our replay (547 / 506 hours respectively) overstate the real-world value once humidity is considered.

If you live in a hot-humid climate, the rule of thumb is strict: only open when dew point is under 65°F. That's a narrow window in summer — typically pre-dawn hours on the coolest mornings.

Where the strategy doesn't work (or barely works)

The data surfaced two bands where the strategy is largely dormant:

Pacific marine — surprisingly limited

We expected Seattle / Portland / London to be peak markets given their mild summer climate. The data says otherwise: afternoons rarely warm above 70°F, overnight lows are often below 60°F. The diurnal swing is small. There's an active summer window (June-September) but the rest of the year is just too cool for our 68-74°F default comfort range to apply.

If you live in the Pacific Northwest or northern Europe and want this strategy to work for you year-round, the best move is to lower your indoor comfort floor — say 62-70°F — and the engine starts catching afternoon-warming patterns that otherwise read as "too cool to bother."

Tropical — mostly dormant by design

Singapore, Bangkok, much of Southeast Asia and Florida — the engine should be telling you "stay sealed" most of the year. Our replay confirms this:

Singapore's 13 open-hours/year tells you the math is brutal. Tropical climates are a control case: if our engine ever starts recommending opens in Singapore en masse, something broke.

How to figure out your own market

If you live in one of the cities above, the data we collected is here: [link to docs/weather_study/climate_marketing_calendar.md or similar].

If your city isn't in our sample, three quick checks:

1. Look at your daily forecast for tomorrow. If the high

and low differ by 20°F or more AND the low is below 70°F, you're in banking territory.

1. Check your dew point at midnight. If it's under

65°F, you'll cool without importing moisture.

1. Look at your AC's runtime over 24 hours. If your AC

runs 8+ hours, there's headroom for natural cooling to shift load. If it runs 2 hours total, you're already spending less than the strategy would save.

The honest truth: ~70% of US households are in a climate that benefits from this strategy at some point in the year. The other 30% should focus on insulation, solar gain control, and HVAC efficiency instead.

The forgetting problem

If natural cooling math applies in your climate, the single biggest failure mode is forgetting — opening windows at 8pm, going to bed, oversleeping, leaving the windows open when you head to work, returning to a 90°F house. Or opening at 7pm before the air has actually cooled down, then giving up.

Most apps that nominally help with this don't actually solve the forgetting problem — they're configuration screens that help you tune your existing habits. WindowWise is the opposite: it watches your local hourly forecast and pings you exactly at the moments when opening will save real money, and again before the heat is coming back. No configuration required beyond a comfort range and a home location.

The core app is free — no ads, no required subscription. Optional Premium add-ons (photographic backdrops, advanced savings dashboards) are one-time purchases. Features whose data has a per-user cost to us (such as detailed pollen forecasting) ship as small optional subscriptions you can turn on if you want; the core windows-timing engine stays free.

[App store + Play store badges]

Where the data comes from

We pulled a year of historical hourly weather (2024) for 30 cities across 8 climate bands from Open-Meteo's free historical API. Then we replayed our engine over every hour, recording the decision it would have made. Aggregated by city by month, the result is a marketing calendar showing when opening windows is worth it where.

The full methodology + per-city data is at [link to docs/weather_study/].