Big data has been revolutionizing the way business and institutions create their strategies, forecasts, and predictions. However, for every big data success story, there are more big data failures. Nate Silver addresses this in his latest book, “The Signal and the Noise.” He dissects scenarios where big data has been essential to forecast predictions across a diverse array of fields like weather, finance, and real estate.
What’s most interesting, is even with terabytes of data, how many of these predictions failed. This happens not only in the fields listed above, but in the world of marketing. As marketers, we have predictive analytics, dashboards, data providers, and other tools which help us gain as much insight as possible and yet, there are times when our efforts completely flop. Why does that happen?
It’s Great When Big Data is Accurate
Meteorologists have been able to greatly increase the accuracy of their predictions over the last two decades. This has happened because they not only understand all the data they collect, they also understand the context behind the data itself. They know exactly how tornadoes form, how tropical storms turn into hurricanes, and are able to collect incredibly accurate data on these events.
Marketers also have access to a plethora of data, and we have the context. With digital channels such as social media, email, display advertising, and paid search we have multiple touch points to analyze. With cross-channel analytics we even know how these ecosystems interact within an integrated marketing strategy. Thanks to first and third party data, we have know our audiences better than ever before. This type of knowledge allows us to use these digital formats and get tremendous scale.
In digital, marketers have control of almost every factor including targeting, channel, copy, and the creative. Yes, the advertising platform algorithms have the final say on when ads are shown. But with all this data, and an in-depth understanding of how the different ad platforms work, marketers can make good prediction on how their campaigns will play out.
But When Big Data Fails
Even with big data, the science of earthquake prediction has made little progress. Seismologists provide a date “range” of when an earthquake will happen, but their official positions is that earthquakes cannot be predicted. Why? Even with all the data that’s been collected, plate tectonics are still not well understood. Thier shifts happen deep inside the earth and making it hard for scientists to observe. The data sets we study and collect represent the results, not the actual movements of the plates.
In marketing PR, organic search, and referral traffic parallel the challenge seismologists face. The data we collect on these channels only represents the results, or the end product. Unlike digital channels, so many factors that determine the outcome of these channels are not easily observed or in the marketers control.
PR is a great example, most marketers know that PR is key in driving great marketing and awareness. The predicament is, even with all the tools and analytics we have in place, we will never be able to create an accurate forecast or prediction on the results of PR. There are just too many factors that are out of a marketers control. For example, it’s hard to predict exactly how a publication will react to a press release, and it’s impossible to predict what other news will get released at the same time. Big data is worthless without accurate data sets, from well understand and repeatable scenarios. Yes, we will be able to formulate a general idea of how PR will perform, but making an accurate prediction is nearly impossible.
Aim for A Happy Medium
With big data, we have the power to take our marketing efforts to new heights, but it can’t solve everything. It’s great for uncovering meaningful patterns in areas of marketing that are well understood and easily measured, like digital advertising or email. However, there are still some areas where big data can’t have a meaningful impact, such as PR. It’s by understanding and working within the limitations of what big data can and cannot do, that it will have the most meaningful impact on your business.