Stefan Bornemann, Head of METEONOMIQS, talks to #machtdigital, an initiative for SMEs, about our weather solutions
Stefan, wetter.com is one of the best-known weather brands in the German-speaking market. Millions of people use your app and website in their daily lives. With your new brand METEONOMIQS, you’ve left the path of B2C business and built a weather-based offering for businesses. What exactly is the idea behind this and what relevance does weather have for business?
To jump right in with your second question: Weather is a very relevant economic factor. This doesn’t just apply to the areas of transport, logistics, agriculture or tourism, which are often the first things people think of when they hear about weather and economy. It also applies to retail, e-commerce, energy, insurance. Ultimately, there is hardly an industry whose success is not also dependent on the weather. Many companies are aware of the weather’s influence, but underestimate it due to a lack of accurate calculations or solutions. This is exactly what we want to change and with our new B2B division METEONOMIQS we offer weather data driven business solutions for different business processes – from demand and order planning to marketing for all channels and performance analysis.
You’re talking about weather data-driven solutions. What exactly do you mean by that and how do they work?
Technically, we offer various API products, i.e. applications via standard interfaces. Economically, we target various use cases, i.e. typical problems.
Via our weather data API, you can easily get different weather data in order to use this data independently in different projects. For a business analysis, a website/app project, a campaign analysis in Google Analytics or for larger data science projects. Here we help with precise weather data for the whole world.
Also easy to integrate is our forecasting tool. With it, we can dock on to common planning systems and improve merchandise, assortment and sales planning by taking the local weather forecast into account.
A further weather application we offer is in the area of advertising. Weather targeting is nothing new per se, but we offer weather triggers that take into account regional differences in the weather. For example, advertising for ice cream at 20 degrees Celsius might work in Hamburg, but not in Munich. That’s why we’ve developed product-weather indices that map customers’ interest in buying a particular product at the daily and zip code level. They can be used to play out online campaigns precisely when the advertising is relevant in terms of location, time, and weather.
Can you give specific examples about how taking weather data into account can make such a difference?
From over 100 projects we have learned the following: The weather effect depends on the industry, the customer, the location and the sales channel. Weather effects of 1% to 10% of sales are common; for beer gardens, day tourism and ski resorts even far above that.
One example: In one of our first projects, we worked with a partner to “weather optimize” the daily ordering process at bakery chains. Here we saw that not only the customer frequency at the counter varies, but also that certain goods sell better depending on the weather. Apple cake, for example, sells better if the weather is good, while sales of eclairs run counter to the temperature trend. So far, these are nice findings that show the influence of the weather in a comprehensible way. However, this effect only becomes economically accessible when weather is systematically taken into account in the sales forecast. By integrating our “weather forecast feature” into existing systems and processes, a forecast optimization in the double-digit percentage range can be achieved. In the area of advertising, we have even observed an 80 percent increase in click rates.
In order for the tools to work, do they also have to be fed with company data, and yes, which data? And are SMEs whose level of digitalization is not yet 100 percent also able to use your solutions?
In general, every customer can use the solutions and does not have to share confidential data. Our solutions use standard tools or are integrated into existing systems via standard interfaces without much effort. For example, we can deliver a “weather add-on” to an existing forecasting/ERP system or set up planning from scratch. For specific applications, such as weather-dependent product forecasting, we rely on enterprise data to make it work. Of course, the data exchange runs under the strictest security and data protection conditions.
Predicting the weather for weeks and months in advance is very difficult and the weather can change abruptly. How far in advance is such a forecasting solution reliable?
Weather forecasts are very reliable nowadays and a lot of work is done to make them even better. As a rule of thumb, without going into meteorological details, we can say that we can forecast well 1-2 weeks in advance. But the question of forecast horizon and forecast quality is often of secondary importance. Many machine learning applications need first and foremost well-prepared historical weather data to separate weather from other effects.
Further information on the initiative: machtdigital.de