Weather data analytics offers transformational insights for retailers

Technology now allows massive weather data streams to  be ‘mashed up’ with other data inputs such as sales data or social media data, and analysed, which in turn helps businesses to understand and predict, at a local level, the precise impact of weather on consumer demand.

For retailers and FMCG companies, the insights derived from the data  when combined with technology integration provide the potential for transformational changes within a retailer or FMCG company says Paul Walsh, Vice President, Weather Analytics and Meteorologist, The Weather Company.

Walsh, who is speaking at the ADMA Global Forum in Sydney in early August, trained as a meteorologist in the US Air Force where he spent 20 years serving across the US and Asia. This included a seven year stint with the US Army.  Indeed despite having worked in startups since 1997, he still took part in the invasion of Iraq during Operation Desert Storm as where he was the Chief Meteorologist for the 101st Airborne Division.

Some description

Interestingly it was his military experience applying weather ‘intelligence’ competencies he learned during 20 years supporting war-fighters to help retailers /CPG companies that helped score him that first startup role

“It was a swords-to-ploughshares transition,“ he says.

Now instead of equipping soldiers with intelligence to fight battles, he and his team at TWC is arming retailers with highly accurate hyper-local forecasts to help them stock their stores and be ready to meet consumer demands.

“Weather has always been a challenge for retailers. The difference now is that weather forecasts are more hyper-local – and yes – accurate. But also those forecasts are accessed by literally billions of people, multiple times daily on their mobile phones.”

Consumer use Weather forecasts specifically for planning, nd that means that each time someone checks the forecast (and imagine a billion people check their forecasts 3 times a day) they are potentially changing their plans, he says.

So with weather being such a ‘variable’ variable, what does the future of weather predictions hold for people in business and retail?

“Integration of weather into supply and demand chain systems is a growing trend that will be accelerating quickly over the next 5 to 10 years,” says Walsh.

“Forecasts are getting increasingly accurate and hyper-local and are being increasingly integrated into systems – soon, all major publicly traded companies will have to have this technology lest they be skewered by the street every time the weather hurts their sales and margins.”

And that ‘variableness’ problem is handled via probabilistic outcomes and learning decision models, he says.

Of course, weather systems are still very different beast to human systems.

Which-50 asked Walsh about the kinds of models TWC uses and and how do they differ from predictive models people might use in a large corporates.

According to Walsh, “We use a proprietary forecast model that combines numerical output from dozens (at least) of predictions as well as other inputs (using an IoT approach) including 130K backyard weather sensors, inputs from sensors on airplanes and, eventually, data from mobile phones and automobiles.”

Additionally,  he said, the company has created a methodology it calls Forecast on Demand. “It literally creates a ‘forecast on the fly’ when you request it based on the latest available inputs.  That data is combined and optimized and the output is (as measured by a third party) the most accurate weather forecast available on the globe.”

He contrasted that to the approach of other forecast providers who simply pull data from the nearest airport reporting station and access a forecast that is updated, for the most part, once every 6 hours.

“For perspective, we produce 15 billion forecasts per day at a global resolution of half a kilometre.”

Social data is also an important part of the mix.”Social data is used in combination with weather data to mine and predict the influence of weather on consumer demand and sentiment – social data is a great external resource. Confidence is gained using standard statistical tests and filters as raw social data tends to be very noisy.”

Sunny, with a hint of transformation

Looking forward, we should expect to see bright days ahead for clever entrepreneurs with new products and business models likely ot be wrapped around weather data.

For example, Walsh mentions an insurance product called WeatherFX Alert.

“It’s a mobile phone app that Mines weather radar data to provide advance warning of large hailstorm that is expected to move over their home or car. It can give users up to 30 minutes notice,” he said.

Paul Walsh is speaking on Wednesday August 5 at the ADMA Global Marketing, Media and Advertising Forum from 11.30am-12.10pm at the Sydney Hilton Hotel. Register to attend at

DIU

Previous post

Is Fin Tech the future of banking?

Next post

Stop hiring data scientists until you are ready for data science