Weather-driven demand and the limits of seasonal inventory planning

Australian retailers in weather-exposed categories commit to summer stock in the middle of winter. Orders placed in July determine what is available in December, and they are placed against a season that cannot yet be described with any confidence.

Pool and spa supplies illustrate the problem in a concentrated form, and the observations below are drawn from Aquatune's experience supplying the category since 1984. The same dynamics apply to cooling, garden, swimwear, sunscreen and outdoor furniture. Standard guidance for these categories is to apply seasonality analysis and layer in weather data. That guidance is sound as far as it goes, but it carries an assumption that does not hold in practice: that each season resembles the last.

Demand tracks deviation rather than absolute temperature

Purchasing behaviour in weather-driven categories correlates less with how hot conditions are than with how far conditions depart from what customers had adjusted to.

The Bureau of Meteorology applies a comparable principle. Its heatwave definition requires unusually high minimum and maximum temperatures sustained across three days or more, and its forecasts account for both what constitutes hot weather in a specific location and the temperatures recorded over the preceding thirty days. The Bureau states the consequence plainly: a higher forecast temperature is required to reach the heatwave threshold at the end of summer than at the beginning, because populations acclimatise.

Retail demand follows a similar pattern. The first sustained hot period of a season typically clears more stock than a hotter period later in summer, because early-season customers have not yet opened pools, tested water or reviewed what remains from the previous year. By late summer those tasks are routine. Conditions are more extreme and purchasing urgency is lower.

Sub-categories follow divergent demand curves

Treating a weather-exposed category as a single forecasting unit produces persistently poor results, because the products within it respond to different signals.

Sanitiser is consumed continuously and depletes faster under heat and heavy bather load, so it tracks conditions closely and sells reactively. Mineral additives sit at the opposite end of the range. A product such as pureswim is dosed at approximately 12kg per 20,000 litres to establish water balance, then topped up with a single bag every six to twelve months. That is a season-opening purchase governed by a maintenance interval rather than by weather, and it is forecastable a year in advance. Equipment and spare parts follow a third pattern, since pumps and filters tend to fail after sustained load rather than during an initial hot spell, placing that demand several weeks behind the conditions that caused it.

Three product groups within one category, following three unrelated curves. Forecasting them collectively produces overstock in one and shortage in another within the same trading week.

Forecast horizons do not align with lead times

The structural constraint is a timing mismatch that inventory software does not resolve.

Long-range seasonal forecasts are probabilistic and revised weekly. Supplier lead times, particularly for imported stock, extend well beyond the point at which those forecasts become reliable. The purchasing decision is therefore made before the forecast carries useful information. The practical response involves less emphasis on prediction than is commonly assumed, and more on structuring the range so that prediction accuracy matters less.

The historical baseline is also shifting. Australia recorded its fourth-warmest year on record in 2025, at 1.23 °C above the long-term average, with maximum temperatures 1.48 °C above average and above-average national temperatures in every month of the year. That followed the second-warmest year on record in 2024. South Australia and Western Australia both recorded their third-warmest years in 2025. Heatwave conditions affected large parts of the country from January to March and again from October to December, and the Bureau's long-range outlook had indicated warmer than average days and nights across most of the country.

Where the underlying climate is trending in one direction, several years of sales history is a weaker predictor than standard averaging assumes.

Practical implications

Recent seasons warrant heavier weighting than older ones. Conventional averaging treats a season from five years ago as equivalent evidence to the most recent one, which understates the trend.

Ranges are better segmented by demand behaviour than by product type. Continuously consumed lines, annual-cycle lines and failure-driven lines each require distinct planning logic and separate safety stock settings.

Stock depth is more efficiently allocated to products that do not expire and are not weather-sensitive, with leaner positions held on lines that spike. Carrying cost on a stable annual-cycle product is low, whereas misjudged stock in a mild season represents the primary exposure.

The first significant heat event of a season is the most informative early signal available. It occurs early enough to influence subsequent ordering and reflects actual customer readiness rather than modelled intent.

Shifting customers from reactive purchasing toward scheduled maintenance routines reduces forecasting difficulty directly. Demand that follows a service interval is predictable in a way that weather-triggered demand is not, and that shift delivers more planning benefit than incremental model improvements.

These measures do not make weather-driven categories predictable. The period from October to April remains Australia's peak season for heatwaves, storms, flooding and bushfires, any of which can reshape a quarter's trading. The objective is not forecast accuracy in isolation, but a range and stock position that remain viable when the forecast proves wrong.

Aquatune has operated from Adelaide since 1984, supplying pool and spa water treatment products and testing equipment, and has advised local and state government on water sanitation and testing for more than 40 years.