Hershey came to us with their business goal: sell 90% of seasonal bags of candy by Halloween. By November 1st, seasonal displays become an eyesore and candy is marked down by 50%.

Despite challenges like slow marketing momentum and a lack of first-party data, Hershey saw an opportunity to use weekly retail sales feeds to guide their marketing efforts.

The data journey began with collecting sales data from retail partner Target, which provided weekly updates on the sales of Hershey's Halloween candy.

Hershey’s custom algorithm then dynamically adjusted ad spend across the Trade Desk, Meta, and YouTube, redistributing advertising budgets based on the latest sales data. The algorithm ensured that advertising was not wasted on areas with great sales, and instead directed ads to places where sales needed a boost.

Data integration was seamless, and the algorithm enabled an agile response to changing market conditions.

Hershey's custom algorithm increased advertising in cities where candy sales were sluggish and scaled back in areas with strong sales. The AI automatically adjusted 676 distinct, zip-level budgets week by week, according to local sell-through vs forecast.

Hershey averaged a 90% sell-through rate of Halloween candy at Target stores, directly attributable to the data-driven adjustments.

Hershey measured a 1.6% incremental sales lift for Target over stores without the custom algorithm

The campaign saw an 80% decrease in data fees.

Hershey achieved 7.6% Halloween Category growth.

Q4 2023 revenue outpaced the previous three years.