Weather moves 35% of GDP. Here is how much it moves your demand.

Ask a marketing lead why last quarter underperformed and you will hear about creative fatigue, a competitor's promo, or the algorithm. Ask about the weather and you get a shrug. That shrug is expensive. Weather is estimated to affect as much as 35% of GDP in industrialised economies, and industry research consistently finds it shaping a large share of consumer behaviour: what people buy, where, when, and in what quantity. Australia is no exception. The Bureau of Meteorology publishes degree-day climatology specifically so the energy, air-conditioning and heating industries can plan against it.

Those are macro numbers. The operator's question is narrower and more useful: how much does weather move my demand, and can I see it coming? The answer depends almost entirely on your vertical, and it is more measurable than most teams assume.

Which verticals are the most weather-sensitive?

The first mistake is treating "weather sensitivity" as one number. It is a distribution. Categories that solve an acute, weather-created problem (a dead furnace, a leaking roof, a hot house) swing violently. Categories where weather merely nudges a discretionary mood swing far less. The chart below ranks peak demand lift in extreme conditions by vertical: HVAC and home services lead at roughly +80% or more; home and garden runs +60-70%; fashion, where weather only shifts the timing of a discretionary purchase, sits near +35-40%.

Why is the signal loudest in search?

For a swing in a number you already have, look at search. Seasonal query volume does not drift, it whipsaws. Peak-to-valley variance for high-intent home-services queries runs into the hundreds of percent.

Spike now, or lag a quarter?

Two verticals can be equally weather-sensitive and still need opposite playbooks, because the timing of the response differs. L.E.K.'s Weather Index for Service Essentials (WISE), a US framework whose mechanics apply just as well to AU verticals, makes the distinction precise.

HVAC lags by roughly one quarter. Degree-days accumulated in a given quarter predict elevated HVAC shipments the following quarter, as heat strains equipment and the replacement cycle lands later. Roofing spikes in the same quarter a storm hits (acute damage demands immediate repair), then shows a larger decline two quarters out as the storm pulls demand forward and exhausts the pipeline (L.E.K. WISE). In Australia the roofing trigger is concrete: the severe-storm and hail season runs roughly September to February in the east (the Brisbane supercell of 24 November 2025 dropped 14 cm hail and cut power to 160,000 homes), and the tropical cyclone season runs November to April across the north (BOM).

That changes where the budget goes. For roofing, you spend into the storm week, which clusters spring to early summer in the east and through the wet season in the north. For HVAC, the degree-days are your early-warning system for a quarter you have not entered yet.

How do you actually measure it?

None of this is proprietary magic. It is textbook applied statistics with public inputs, built on two blocks: degree-days and lagged correlation.

A degree-day measures how far a day's temperature sits from a comfort baseline. The Bureau of Meteorology publishes Australia's heating and cooling degree-day climatology against a base of 18 °C, the metric degree-day base (BOM). Above it, cooling demand accrues; below it, heating. The geography reverses the northern hemisphere: cooling degree-days rise toward the north (Brisbane averages roughly 1,000 a year) and heating degree-days toward the south (Melbourne only about 250 cooling, far more heating).

To test whether that demand actually tracks weather, the defensible tool is a Pearson lag correlation: shift the weather series forward by k periods and measure how strongly it co-moves with demand. The lag that maximises the coefficient is your lead time.

The trap that makes weather look bigger than it is

Here is where most naive analyses overclaim. Correlate raw demand against raw temperature and you will flag almost every seasonal category as "weather-sensitive," because both follow the calendar. The fix is to control for climate (the predictable baseline seasonality) and isolate the residual that genuine weather deviations explain. Skip this step and you attribute ordinary seasonality to weather (ResearchGate, 2023). The honest method uses multiple linear regression with autoregressive terms, netting out climate first.

Done properly, the payoff is real but modest and varies by category. Adding weather to a demand forecast improved accuracy by +20.2% for grocers, +12.2% for casual dining, and +3.3% for home improvement in one multi-sector machine-learning study ("The Substantial Role of Weather Data in Consumer Spending Prediction", 2024, US data). Anyone promising a step-change from a single weather feature is selling you the uncontrolled version.

How far ahead can you actually see it?

A forecast is only useful if it arrives before you need to act. The honest horizons:

Horizons per WebFX seasonal guidance. The practical rule that falls out: launch 30-45 days before peak. In Australia that means cooling campaigns are live by October ahead of the Dec-Feb summer, and heating campaigns by April ahead of the Jun-Aug winter.

The two ends of that table do different jobs. The seasonal pattern tells you the shape of the year and when to be ready; the short-range forecast tells you when to pull the trigger. A team with only the second is always reacting; a team with both is positioning ahead of demand.

What does this mean for the budget?

The conclusion is not "spend more in summer." It is sharper. When demand shifts between seasonal service types, pivot spend toward where demand is, do not cut. A high CPA in a trough is not inefficiency to be slashed; it is a signal that demand has moved to a different service line. Cutting budget in the trough and scrambling to rebuild it at peak is the most common and most expensive mistake in weather-sensitive verticals.

What to take from this

• Rank your sensitivity honestly. A furnace-replacement business and a fashion brand do not share a playbook; amplitude and event frequency both matter.
• Know your timing. Same-quarter spike (roofing) versus one-quarter lag (HVAC) decides whether degree-days are a trigger or an early warning.
• Control for climate. Net out baseline seasonality before crediting weather, or you will overclaim and over-spend.
• Position 30-45 days out. Seasonal patterns are forecastable months ahead; the budget should already be shaped before peak arrives.

Weather is not the excuse for a soft quarter. It is one of the few demand signals you can see coming.

The operators who win the season are the ones who modelled it before it landed.