Making data “big” is not so hard, really. Much of that work has been done for us by the digital revolution itself. The problem of finding the truly meaningful points within the bottomless fit of data our interactions cast off has been with us since the earliest Web site analytics packages rendered bigger stacks of accordion-format print-outs than most publishers cared to peruse. In fact the current fashion around “big data” is a bit of a misnomer. Data has been “big” all along. What has changed now is not just scale and cross-channel inputs, but the sheer speed and accessibility of data as it moves to the cloud and becomes present on any device anywhere. Making data actionable in real time and at the point of critical need or decision-making is where data is not just big, but enormously effective.
As we prepare for Mediapost’s first OMMA show devoted to “Data Driven Marketing,” I have been exploring with a number of executives how they are using these emerging tools to conceive new products and services. Few are as interesting as The Weather Channel’s use of its own mountain of weather and user data to craft new business models for itself. In many ways they are charting a new path for media companies who are facing the prospect of seeing ever more of their content commoditized. “Media is not a standalone currency without the judicious application of data,” says Vikram Somaya, General Manager WeatherFX.
His group within Weather channel is not just using data to increase media value in the usual ways – tighter targeting, granular segmentation, high CPMs, etc. “We have a leverage-able proprietary asset, not just weather data, but analytics we put on top of it and linkages we take from that weather. We create data products that take weather data that has been a commodity and build intelligent analytics for marketers.” The products are aimed at industries that are most obviously affected by weather, from retail to insurance to quick service restaurants. The idea is that these categories benefit not only from knowing the weather that relates to their business and product cycles but from knowing how purchase patterns change historically as a result of weather.
WeatherFX is working with an insurance company, for instance, to monitor especially threatening weather events like imminent hail, which produces property damage. Within a half hour of hail reaching a specific region, SMS alerts from the insurer go out to customers suggesting they bring their cars into garages.
WeatherFX is developing APIs that let this combination of real-time and historical data and analytics tie into touchpoints like point of sale. By empowering the cash register to know how purchase patterns change in relation to approaching weather, POS within a geofenced area can now be triggered by real-time weather to make special promotions or signal reminders to cashiers.
“We are building the data back end to link three pieces of weather data that influence commercial decisions,” says Somaya, “forecasts, real-time data and historical data. All of that is useful when creating models for commercial behavior.” But if APIs are going to bring that intelligence and triggering mechanism down to actionable points like cash registers and mobile phones, it is just as important that the new machine has a feedback loop in order to learn from and optimize against performance. Somaya emphasizes that these relationships also require a higher level of data sharing among the partners. “One requirement for this working is sharing of sales and revenue information. A part of what we are developing is a performance modeling piece with machine learning that folds back into the libraries. That is the most important component to me in validating the value of this.”
For other media companies, WeatherFX suggests how data can be leveraged in multiple ways to craft new products that are aimed at new constituencies. This is a model that stretches the idea of the traditional media company, which built and sold space and audiences. The media company is not just selling content, context or even audiences or audience intelligence. In this model, content and its consumption by audiences combined render proprietary information that becomes its own wholly different product altogether. Data becomes the new kind of media a publisher produces.