Why data science is failing marketers

Companies can now gather more information about their customers than ever before. But according to a new study data science is not benefiting marketers, with 84 percent of marketing executives saying their ability to predict customer behavior is guesswork.

The report from predictive analytics company Pecan AI, based on surveys carried out by Wakefield Research, finds four out of five marketing execs report difficulty in making data-driven decisions despite all of the consumer data at their disposal.

The study of 250 respondents finds 95 percent of companies now integrate AI-powered predictive analytics into their marketing strategy, including 44 percent who say that they've integrated AI-powered predictive analytics into their strategy completely. But among those whose companies have completely integrated AI predictive analytics into their marketing strategy, 90 percent report that it is difficult for them to make day-to-day data-driven decisions.

All respondents say they want to gain additional AI-powered capabilities and predictive insights for their teams, clearly indicating that current implementations of predictive analytics are poorly serving the needs of today's marketing teams.

"With most companies today employing manual model building approaches, it's unfortunate, but not surprising that the results are failing the needs of marketing teams," says Zohar Bronfman, co-founder and CEO of Pecan. "While data scientists may be skilled in building the perfect software models, they are simply too far removed from the nuanced realities of the business to be effective. In addition, given their workloads they are too slow to respond when considering the rapidly changing market conditions and consumer behavior. Marketers and marketing analysts are more than capable of handling predictive analytics responsibilities if provided with the right tools."

So what's holding back the success of data science projects? 42 percent say data scientists don't have the time to meet requests, 40 percent say those building the models don't understand marketing goals, 38 percent say data scientists don't ask the right questions and 37 percent say that wrong or partial data is used to build models.

In addition 93 percent of marketing executives polled agreed that data scientists could solve more complex problems if they were able to use low/no code AI predictive modeling tools for automatable metrics.

You can get the full report from the Pecan site.

Image Credit: Syda Productions/Shutterstock