The capability is the result of a project conducted with the start-up Simplex, which integrated the history of three years of 10 different databases - ranging from sales results over time, investments made in different media, product participation in points of sale, retail prices, and demand forecasts, to the regional advance rates of the Covid-19 pandemic. The project went further. Another of its pillars was developing a tool to automate the application of statistical models on this sea of data to enable scenario forecasting.
"Today, we are not only able to measure the results that TV, radio, or digital media campaigns have had on sales, but we already have an algorithm that predicts, with 75% accuracy, what the ROI (return on investment) of future campaigns will be in each region of the country," celebrates João Galoppi, head of Digital Marketing at Bunge.
According to the executive, the accuracy of the predictions tends to increase since we improved the statistical predictive models adopted through artificial intelligence. For him, one of his biggest challenges was understanding why a campaign worked so well in one place, not another. And, of course, whether the marketing budget was well-dosed among the various media available.
"Obviously, we had some assumptions. But today, on the contrary, we have organized data and the innumerous possible crossings", he highlights.
According to the executive, Bunge today can already identify, in detail, the weight of the price increase in a specific market in neutralizing the marketing efforts destined for that region. Or to see the peculiarity of a particular market and how its consumer or buyer reacts - since they are not always the same person - as a result of promotions or advertisements.
Galoppi argues that optimizing the investment based on data is the most efficient way to overcome the crisis and increase the marketing budget's potential reach, especially in the current scenario. "Simplex was decisive in this process given its expertise in predictive statistical models and the development of the tool that allows us to see the whole scenario through intuitive dashboards," he summarizes.
Data under the magnifying glass
The integration and cross-referencing of information was done on 10 different databases, considering both the advertising and marketing databases and the databases for physical sales channels. Among the various information gathered and organized, Galoppi enumerates the audience points of the campaigns carried out, the cost of online and offline campaigns, the participation of each type of campaign, and the total number of impressions and clicks. On another front, Nielsen data were also surveyed, such as the brand's volume and share on retail shelves, the months in which Bunge advertised, and the periods in which it carried out marketing actions aimed at the offline sales channels.
"What we have in our hands today is better-grounded decision-making," guarantees Galoppi. For the person responsible for Digital Marketing at Bunge, Simplex's participation in the process was decisive: the start-up allowed more speed to the work developed. "We concluded our plan in less than 6 months", he celebrates. For the executive, another success factor of the initiative was the work of a multidisciplinary team, including a dozen people, from advertising professionals to chemical engineers, and involving several areas of the company, which gave a richer and more diversified view of the challenges posed.
"In digital markets, everything can be measured, from the first click to the conclusion of the purchase; it is possible to invest in media with precision and calculate your ROI (Return Over Investment). And a great challenge, achieved here, was to do the same in offline media effectively", explains João Lee, director of Simplex. He says his greatest satisfaction with the project is seeing that algorithms and data already support several other company areas to better understand market behavior.