Weather data analytics: Helping retailers predict and meet customer demand
As an old retailing saying has it, success is all about “location, location, location.” But you could just as easily argue that success is also about “precipitation, precipitation, precipitation.”
When the rain falls and the wind howls, shoppers may cancel or postpone shopping trips to avoid braving the elements. Accordingly, atmospheric conditions can mean the difference between a robust shopping season and one that falls short of expectations. Moreover, weather can disrupt suppliers, shippers and other components of the retailing value chain, making it hard to serve customers who came out despite the weather. Weather can even affect business by triggering subtle shifts in customer sentiment, as when prolonged heat waves increase general levels of listlessness and fatigue.
Examples of weather-related influences on consumer demand abound, says The Weather Company:
- Sales of supermarket ice cream dip when temperatures exceed 25°C (77°F)—shoppers fear that it will melt.
- Barbecue sales triple in Scotland when temperatures rise above 20°C (68°F). In London, however, the figure is 24°C (75°F).
- Periods of hot, sunny weather improve people’s moods, producing better reception to advertising.
Bring efficiency to the planning process through meteorology
Clearly, meteorology can be just as important to retail success as geography is. Weather patterns influence whether customers choose to visit brick-and-mortar outlets rather than to shop online—another reason why retailers need to continually adjust their operations to mitigate business risks associated with storms, seasons and other weather patterns. Retailers of all types can use weather data analytics to aid evidence-driven decision making. Indeed, the insights gained from analysis can be as much business opportunity as they are risk mitigation strategy.
For example, retail businesses can incorporate weather information to bring efficiency into the planning process and to deepen customer engagement. Historical analysis, forecasting, predictive modeling and geospatial visualization, among other statistical analysis tools, can help address challenges in more than one area:
- Supply and demand planning
Retailers can use both long-term and short-term weather data to forecast supply and demand for specific locations as a way of optimizing basket mix and inventory. Historical data are useful in calculating weather’s effects on store traffic, seasonal inventory levels and customer market basket mixes.
- Marketing and personalization effectiveness
Retailers can create efficiency in personalized marketing campaigns by using local weather data—and not just to boost the effectiveness of advertising and marketing campaigns for seasonal products. Much as hyperlocal weather data and analytics can shed light on times, areas and store locations that can be used to heighten marketing campaigns’ effectiveness, weather data and social and customer buying analytics are useful for identifying what products to offer when.
In one example of this, IBM helped a leading coffee retailer use weather data to trigger its marketing campaigns. The client needed deep customer purchase behavior insights at a store level to step up marketing effectiveness, drive transactions and boost ticket size, thereby enhancing customer loyalty. As the figure shows, IBM developed a pattern detection engine to detect and classify customer purchase patterns driven by weather—among other factors, both internal and external—at the levels of individual customers, stores and products. Accordingly, thanks to targeted actions driven by weather-related behaviors, the client saw a $44 million increase in incremental marketing opportunities.
From a risk-mitigation perspective, retailers can use weather data analysis to address many challenges:
- Predicting weather’s adverse effect on consumer demand
Weather affects store traffic very evidently—particularly when a blizzard or other major storm disrupts customers’ normal shopping routines. But weather also affects in less obvious ways the products that customers buy. For example, a hot, humid summer can boost demand for soft drinks while dampening demand for jackets and back-to-school clothing. By using advanced analytics to process historical, current and forecasted weather data, retailers can predict how shifting demands will affect inventory, merchandising, marketing, staffing, logistics and more.
- Ensuring inventories sufficient to meet weather-driven demand
Retailers who don’t stock specific items during critical opportunity windows may be out of luck. Weather events can suddenly affect inventory levels for particular products, as when blizzards trigger a run on snow shovels, rock salt, canned goods, toilet paper and other such items. However, although weather-driven inventory management challenges such as this may be predictable according to an annual regional cycle, the timing and severity of specific events can be difficult to pinpoint. Retailers who can’t predict effects down to the ZIP code level, determining which stores should stock up on which items to respond effectively to coming adverse weather, may be blindsided, running short not only on important products, but also on available staff members.
- Mitigating weather-related disruptions to supply chains
In retailing as much as in any other industry, supply chains must be sustainable. Weather events that are of long duration can adversely affect retailing by disrupting established sources of supply. For example, a lengthy drought in California might disrupt regional agriculture, forcing grocers to find alternate suppliers—whether domestic or overseas—for many produce items. Likewise, especially intense hurricane seasons heighten the risk of flooding in affected regions, threatening a wide range of industries in those areas. Retailers need statistical, geospatial and predictive tools to help them model various risk factors and identify alternative sourcing plans to mitigate likely adverse scenarios.
Retailing stakeholders who stand to benefit from data-driven weather analysis include chief merchandising managers, purchasing officers, supply chain managers and store managers. Analysis should incorporate both the retailer’s own transactional data (sales, inventories and the like) and weather data enabling comprehensive modeling of relevant scenarios. Such data may include third-party information (historical, real-time, forecasted) about severe weather, seasonal patterns, air quality, pollutants, and pollen and other allergens and should include data derived from remote sensing earth imagery.
Learn how to do your own weather-driven analysis by attending IBM Insight 2015, scheduled for 25–29 October in Las Vegas. Together, IBM and The Weather Company will demonstrate how to use The Weather Company weather data packages and IBM data science tools—including Apache Spark as a Service in the Bluemix cloud—to build powerful smart data applications for meteorological forecasting to help you address urgent weather-related business challenges.
Discover how everyday weather affects retail, then learn more about how to use Apache Spark as a Service to hack your analytics-related business challenges.