You’re invited to AW2020, Advertising Week’s digital event, September 29-October 8 to help work through solutions to some of the advertising and marketing industry’s biggest problems. From climbing unemployment to racial inequality and an unclear future, now is the time, more than ever, to think and work together. Register to learn more.
Predictive analytics, the use of real-world data and machine learning techniques to identify the likelihood of future outcomes, has become a cornerstone of media planning. Human intelligence, however, is just as important to the success of predictive analytics’ application as technology is. To understand how, we spoke with Piotr Urbanski, Ayzenberg associate director of marketing science, and Ayzenberg chief marketing officer Vincent Juarez, who both work closely to craft client strategies based on historical data and predictive analytics, or predictive modeling. The two discuss all things predictive analytics: what it is, why it’s important for marketers and why it’s critical for media planning, especially during times of uncertainty.
What are predictive analytics?
Piotr Urbanski: There are two approaches to analyzing data—the first is predictive analytics or predictive modeling, And the second is causal inference, which is a little bit more descriptive and aims to understand the cause and effect between the numbers. For example, if one goes up, why does the other one trend with it? Whereas predictive modeling doesn’t consider the why; you just want to be as accurate as possible. Those who use predictive modeling know it’s a very good predictor and they get very good outcomes, but they don’t measure the why.
How do predictive modeling and causal inference differ?
PU: They use the same statistical tools. The most basic predictive models can perform a regression run or trend line. As one value goes up, the other value also goes up. And then you could extrapolate, or predict, based on that trend.
Causal inference takes it a step further to consider theoretically how A impacts B. For example, if A is temperature and B is the crime rate, you see that as temperature goes up, the crime rate also goes up. Temperature is a good predictor of crime. But does that mean temperature causes crime? So at that point, you start digging into various different theories.
For analysts, causal inference makes our job much easier. Because part of your research is already researching the cause and effect, you get really easy, actionable insights. So in most cases, now that you’ve identified a cause, you can control that cause or influence it in some way. In a nutshell, predictive analytics and causal inference are two sides of the same coin.
Why are predictive analytics important for marketers?
PU: On the media side, predictive analytics helps predict where you should be targeting money by analyzing the behavior of consumers purchasing your product or engaging with your brand. To be really good at predictive analytics, a human element is still necessary. So on top of your findings, you’ll typically have a marketing science team that layers additional analysis to not only make sense of the data but also predict between platforms to achieve a more wide-reaching view.
At the end of the day, predictive analytics help a company’s bottom line. Let’s say you’re running a campaign that’s a million dollars. If you can become more accurate by even a fraction of a percent, you’ve saved tons and tons of money and easily added value that wouldn’t have existed before. It also makes your actions accountable because you know what isn’t working and where to stop or start paying for ads to get the biggest bang for our buck.
How do predictive analytics impact Ayzenberg’s media department and planning?
Vincent Juarez: I oversee both media groups at Ayzenberg, the media planning group and the influencer marketing group. A lot of our clients rely on us to provide them with maximum transparency in the campaign strategy that we craft for them. This is not only to validate the results of the campaign, but also to create models that predict what those results are going to be. That’s where predictive analytics become especially important for marketers because in this day and age, we’re held to a higher standard of accountability. As a result, the agencies that work with them are held to an even higher standard of accountability to try to predict a certain amount of return on your investment.
The relationship between our media department and marketing science team is symbiotic. We provide the wealth of data that gets fed into the marketing science team’s models to try and predict outcomes based on different marketing mixes, different marketing strategies and different budget allocations by vehicles and so on.
It’s hard for Piotr and his team to do their job without the data that my team generates around the real world basis. He takes our analytics and our forecasting to a scientific level that our media planners and marketers just don’t have.
This is one of the reasons why after many years, we identified the need for a dedicated marketing science group. As a digital- and social-first business model, we found that there was a distinct need for a marketing science group to provide us with that next level of expertise to assess the health of a campaign and predict the outcomes of campaigns that we create on behalf of clients.
How has COVID-19 affected the media team’s use of predictive analytics?
VJ: Because of the pandemic, marketers are a lot more careful about their investments. There’s a lot of scrutiny in terms of making every dollar work like ten dollars. As a result, predictive analytics has become even more important during this time.
Though we don’t have control over external factors that could potentially affect our forecasts, I’d say we’re very successful in our predictions. We have made a significant investment in data science because everyone wants to go into their strategy with as much information as possible to determine the success rate.
Predictive analytics offers a safety net for marketers and for agencies to try and understand upfront the degrees of success or the risk of running a certain type of campaign. Instead of the old style of throwing millions of dollars at a specific vehicle like TV, for instance, and watching the sales numbers and hoping for the best, it’s really about trying to predict the future of the cause and effect of investment. And having that kind of knowledge helps us differentiate ourselves from other agencies. This article first appeared on a.list.