With the IoT, Data Scientists Must Become Data Creators

It's No Longer About Collecting Data, but Determining How It's Created

By Published on .

Reprints Reprints

A Samsung Smart Things door lock.
A Samsung Smart Things door lock. Credit: Chris Ratcliffe/Bloomberg

The proliferation of the internet of things allows us to see that every single action and reaction in the world around us (and even within us) has the capacity to be measured. The revolution is still in the works, and marketers are closely watching how it will impact the way businesses are managed, campaigns are developed and, of course, the way agencies work with data.

In his HBR article "Design Thinking Comes of Age," Jon Kolko said "the business environment is so volatile that a company must experiment with multiple paths in order to survive." If businesses are becoming more volatile, then the data collected to evaluate each of these paths needs to be part of the initial design in order to allow these paths to be properly evaluated, optimized and, in some cases, enabled by leveraging the right IoT solutions.

Data scientists have to add this new layer of complexity to their work. Instead of waiting to receive a .csv file or to download data from a Hadoop cluster, he or she will have to think through how the data should be created in first place. Therefore, the concept of the data acquisition process shifts from "where is the data?" to one that allows for the data to be created, or improved upon in cases where it is not yet available or satisfactory.

The idea of tracking external factors and understanding how they impact behavior is not new, and its limitations have already been demonstrated by the Hawthorne effect – a phenomenon in which individuals modify their behavior in response to being observed. However, the ability to create new and specific data gives us opportunities to capture streams that can help us identify new signals in the noise.

Retailers want to increase sales, and if they provide valuable information while pushing their product message, it will make these interactions between customers and retailers immensely improved. This new framework requires new tools, and data scientists will have to adapt to new and improved software and IoT solutions like Arduino and Intel Galileo. These IoT solutions can continually collect different types of information as well as interact with other IoT devices and send real-time information to a central server.

For example, auto companies and dealers in some cases still rely on manual reports that count "ups" -- i.e. the number customers visiting a store -- so why not start designing new ways to collect information that can improve foot-traffic information and customer interactions?

Let's say we want to know which car model people spent more time looking at. These new IoT tools allow data scientists and marketers to brainstorm ways to capture customer interactions with a retailer. One option could be installing sensors in each of the cars in the showroom to measure specific interactions by model, and another might be to use cameras that capture facial expressions of customers test-driving the vehicles.

So, we would go from knowing that 50-plus customers came into the dealership today to knowing that 50-plus people interacted with five out of eight models; the most-visited model was X; and the most-enjoyed during test drives was Y. Now correlate this data against sales and ad spend, and you have an improved response model that can help you optimize your marketing strategy. Of course, there are many obstacles to take this idea from a prototype to a fully operational and scalable solution, but thinking about the problem with these new tools can only help.

The rise of IoT solutions gives us endless alternatives to understand and interact with the world around us. This is not to say that IoT solutions will replace other methods, but rather work together with existing ones. IoT will also give companies new responsibilities because of the sensitivity around data privacy, but from now on, data scientists and marketers will need to evolve from data disciples to data gods.

Most Popular