• by Laurie Sullivan  @lauriesullivan, November 18, 2020

“Don’t use small data to answer a big-data question,” said Zach Baze, an expert data scientist and chief intelligence officer at agencies Hawkeye and Leo Burnett.

He believes determining the outcome and serving consumers the correct creatives requires listening and analyzing what has been said when people don’t think they are being asked the questions on platforms from search and social data, to secondary and primary research. Look at what is being shared.

Baze, who oversees data design, research and analytics, calls “signal data” the triggers.

These are moments in search and on social platforms like Reddit, Instagram, TikTok, Facebook, or other places were unstructured data lives -- places where people don’t think they are being watched. Free search tools like SEMrush can be used to analyze trends and compare it with primary research.

Marketers need to expand on the ways they predict results.

Marketers don't rely on surveys to predict sales results, Baze said -- they analyze historic information and build models. That’s the strategy behind agencies Hawkeye and Leo Burnett.

Marketers should apply machine-learning algorithms to look at sentiment and conversational volume and evolution. With the correct machine learning applied, marketers can tie conversations to an individual and determine whether that individuals’ conversation has remained consistent or evolved over time, as well as the conversations that the original conversation changed or influenced.

“You tend to get a more truthful understanding of their behavior and attitude,” he said.   

Pollsters and pundits continue to misunderstand data. Baze said marketers should continue to poll away, but combine signal data to get to more reliable outcomes. He believes that this approach allows the advertising industry to better understand and anticipate behavior — whether it’s casting ballots or trying to reach consumers who buy online.

Polling and data science are implicitly different. With polling, marketers only focus on an outcome and a narrow issues, whereas Big Data is about all the context.

“Relying solely on polling creates a blind spot,” he said. “You’re relying on what people tell you. I have two kinds, and when I question the answer, I rely on a third source.”