Digital promotion continues to evolve through platform companies' data utilization. Maximizing Effectiveness with "SP COMPASS" (Part 1)

Dentsu Inc.and Dentsu Digital Inc. have released " SP COMPASS " (February 2023) as a jointly developed new solution. This system simulates participant characteristics and optimal campaign conditions for digital promotions.

Historically, consumer participation in campaigns primarily relied on analog methods like collecting stickers and mailing them in. However, recent years have seen a shift toward digital methods, such as smartphone-based entries. We interviewed Dentsu Inc.'s Masahiro Fukuda and Dentsu Digital Inc.'s Kanon Iino, who were involved in developing "SP COMPASS," to explore the background of its creation using real-world examples.

Learning from past data and predicting effectiveness through algorithms

Q. First, could you tell us about your usual work responsibilities?

Fukuda: My work primarily involves analyzing advertising and sales promotion strategies using data clean rooms held by platform companies on the internet. I also support internal system development.

Iino: I belong to the Commerce Division at Dentsu Digital Inc. My job is to provide solutions that support consumers at the "moment of purchase." This involves planning using digital platform companies' sales promotion solutions, particularly in the offline space, as well as online mall e-commerce and proprietary e-commerce. The "SP COMPASS" project was part of this work.

Q. According to the release, "SP COMPASS" is a system that enables participant prediction and simulation of optimal campaign conditions for digital promotions. Could you elaborate a bit more on the system's functionality?

Fukuda: "SP COMPASS" is a system that analyzes past campaign data using a proprietary algorithm powered by machine learning to simulate the effectiveness of promotional campaigns.

Specifically, when you input information about a planned campaign initiative—such as the campaign period, target distribution channels, and the number or amount of products required for participation—it calculates the estimated number of participants (≈ campaign effectiveness) based on past performance. Using the same mechanism, it can also derive campaign conditions that maximize participant numbers.

Q. I see. Both Dentsu Inc. and Dentsu Digital Inc. were involved in developing this service, right? How were roles divided for this collaboration?

Iino: Within the Commerce Division I belong to, we have many planners who daily propose solutions combining ideas, technology, and data to various companies. Dentsu Digital Inc.'s role was to gather feedback from the field—such as "We need something like this" or "Can't we do it more like this?"—and then propose and drive the system development.

Fukuda: So, the Dentsu Inc. side developed the system based on those proposals. We built it together through constant information and opinion exchange.

Leveraging the Shift in Sales Promotion Methods from Analog to Digital

Q. So this service heavily reflects needs from the field. What was the original impetus for development? Could you share the background?

Iino: The foundation was a campaign strategy using PayPay Gift, established about two years ago across the entire Dentsu Group. Previously, consumers typically participated in corporate campaigns by collecting receipts or stickers as proof of purchase. However, cashless payment services like PayPay retain purchase records, meaning simply buying with PayPay automatically enters you into the campaign. This represented a major shift in the promotional landscape.

However, unlike traditional campaigns we've run multiple times, such as postcard entries, "digital promotions" like PayPay lack prior experience, and results remain unpredictable until launched. This makes it difficult to take the plunge. Amidst this, the question arose: "How can we update campaigns using PayPay Gift?" Specifically, focusing on three aspects—the entry mechanism, point incentives, and promotional volume—where should budget be concentrated to maximize entries? This was a challenge not only for client companies but also for Dentsu Inc., highlighting the need for campaign simulations. This is how the concept for "SP COMPASS" was born.

Campaign participation methods have diversified—collecting stickers, scanning QR codes to enter, etc.—but the number of participants and the demographics involved vary significantly for each method. Discussions arose about how great it would be to avoid making this experience dependent on specific individuals, to properly accumulate that data, and to use it for predictions. This led us to move forward with system development.

Q. So, you both had a vision combining the evolving campaign methods with the needs of client companies and within the company itself, and you translated that into a service.

Iino: That's correct. As planners at Dentsu Inc. proposing campaign strategies, while we can formulate hypotheses based on past experience, whether they are correct or applicable this time remains unknown at the proposal stage. By utilizing "SP COMPASS," we can preemptively understand trends in how participant characteristics change based on point incentive rates, gaining insights for planning. By simulating which of multiple designs is likely to maximize participant numbers, we feel this approach can further increase the contribution to client companies.

For instance, we've frequently heard from client companies wanting to know which approach yields better results: granting 50 points (equivalent to 50 yen) to every applicant, or awarding 2000 points (equivalent to 2000 yen) to winners selected by lottery. Naturally, the cost-effectiveness of promotional activities is also crucial.

Fukuda: A key feature of SP COMPASS is that, beyond simulating upcoming campaigns, it allows us to understand "if we run a campaign designed this way, these types of people are likely to participate." By targeting these "likely participants," we can effectively promote the campaign. Furthermore, data visualization reveals whether those deemed "likely to participate" align with the client's original target demographic or represent a new segment. This capability addresses needs like "wanting to attract new demographics." Note that "SP COMPASS" uses only data with fully verified user consent and conducts analysis within a privacy-protected environment called a data clean room.

The accuracy of predictions made by machine learning models increases with the amount of historical campaign data available. It is expected that the more this system is used, the more data will accumulate, making it easier to leverage for future campaigns.

Just as many services are digitizing, companies' campaign strategies will also increasingly shift to digital. In this environment, the key to effective strategies seems to lie in how to leverage the data that can be accumulated precisely because of digital initiatives. How to utilize customer data and connect it to subsequent initiatives to maximize effectiveness—SP COMPASS addresses these challenges.

In the second part, we will delve deeper into the practical application of "SP COMPASS" while also discussing the significance of digital sales promotion and future service developments.