Each year, America's largest retailer generates about $14-15 billion in marketing-related sales, which means that decisions about how to allocate marketing funds—and media spending in particular—are not taken lightly. But using historical data to decide what to spend money on among the dozens of available channels, from traditional TV to Tik Tok, is not easy. Data is often out of date by the time it can be analyzed, and new channels and platforms are constantly on the rise. With so much money at stake and quick answers hard to come by, increased speed and flexibility were high on the retailer's wish list, and the company set out to get more specific and actionable information faster. When technology meets human ingenuity partnered with a retailer to develop an AI-based solution that will enable faster and better data collection and more accurate modeling to optimize media spending. The first task was to accelerate the existing data flow process and then aggregate and process all the data from media channels, sales, and expenses that were used in the measurement model. By setting up AIP+, pre-integrated services and AI capabilities, for data aggregation, we helped reduce the existing process by 80%, using automation to speed up processing and validation.

After considering the data flow, the team decided to change the underlying model that produced the measurement. Previously, these models were based on hypotheses, i.e. people carefully hypothesized about every possible interdependence between different channels. New machine learning has been introduced into the process to help proactively identify those interdependencies between channels that potentially drive sales. With the new monthly cadence, the team could update the models each month, repeating the previous month's model, rather than starting from scratch. By conducting in-depth training sessions for employees on modeling methodology, the team offered them transparency that generated approval and credibility for the solution.

The number of marketing channels included in the simulation was increased by nearly 40%, allowing them to thinly slice the data (for example, splitting all "social media" channels into each social media platform). Valuable difference The results were significant. The solution reduced the delay between the measurement period and performance analysis from five months to five weeks, opening up 10 and a half months of planning for the same period next year. In addition, moving from a single annual measurement (where performance was expressed as an average) to monthly measurements meant that the understanding was more granular, so the team could see how the performance of a particular channel might change over the course of a year. More specifically, the team estimates that $300 million worth of media buying and value creation opportunities have been opened up through the introduction of the new tool. This meant the team could spend the same amount on media and increase sales by $300 million. .