In-store one-to-one marketing

Diego Klabjan, Jin-Xiang PeiJournal of Retailing and Consumer Services 2010 · DOI

TL;DR

This paper introduces a novel approach to in-store one-to-one marketing by dynamically guiding shoppers through a retail store using personalized routes and targeted promotions. By leveraging technologies like mobile devices and RFID, the authors demonstrate a significant potential for increasing store revenue through induced impulse purchases.

What Problem Does This Paper Address?

Traditional one-to-one marketing efforts in retail, such as targeted emails or personalized website recommendations, primarily occur before or after the shopping experience. The crucial period during the shopping trip, where a significant portion of purchasing decisions are made, has remained largely untapped for personalized marketing. This paper addresses this gap by proposing a system that intervenes during the shopping experience to influence customer behavior and drive sales, particularly through impulse purchases. The challenge lies in effectively identifying opportunities for personalized promotions and guiding customers to those promotions without negatively impacting their shopping experience.

Methods

The paper proposes two main technological scenarios for implementing in-store one-to-one marketing:

  1. Mobile Device and Shopping List Scenario: Customers use a personal digital assistant (PDA) or smartphone with a preloaded shopping list. This list is transmitted to the store's system, which then computes an optimized shopping route. This route is designed to include locations with promoted items, leveraging historical purchasing data and customer preferences to maximize the store's expected revenue. This scenario utilizes discrete choice models and a generalized orienteering problem framework to determine the optimal route.
  2. RFID and Smart Shelves Scenario: This more advanced scenario involves RFID tags on items and smart shelves equipped with interrogators. This allows for real-time tracking of items in a customer's shopping cart. Based on the current basket contents and promotions, the system can issue targeted coupons and recommend a route to redeem them. This scenario incorporates market basket analysis to better predict impulse purchase probabilities.

Both scenarios rely on discrete choice models to predict customer purchasing probabilities and a generalized orienteering problem formulation to optimize the shopping route for maximum store revenue, while also considering customer utility and time constraints.

Datasets

The computational experiments were conducted using real-world data from a major grocery store. This data included:

  • Store layout and complete planogram.
  • Distances between categories.
  • Point-of-sale (POS) data from representative customers (averaging 30-50 items per trip, across 7 shopping trips per customer).

Promotions were randomly generated to simulate various levels of promotional activity (1%, 3%, and 6% of products).

Key Insights and Ideas

  • Marketing During the Shopping Trip is Key: The most significant insight is that personalized marketing can be highly effective during the shopping experience, not just before or after. This is where a substantial portion of purchasing decisions are made.
  • Route Optimization for Revenue: By intelligently routing customers, stores can strategically expose them to promotions, thereby increasing the likelihood of impulse purchases and overall revenue.
  • Technology as an Enabler: The paper highlights how emerging technologies like PDAs, smartphones, and RFID are crucial for enabling these in-store personalized marketing strategies.
  • Balancing Revenue and Customer Experience: The models incorporate customer utility and time constraints, acknowledging that an overly aggressive revenue maximization strategy could negatively impact the customer experience.
  • Promotional Thresholds Matter: The study suggests that the revenue impact of personalized promotions is relatively flat up to a certain threshold, beyond which it increases substantially. This implies a strategic point for store managers to focus their promotional efforts.

Key Results

  • Substantial Revenue Increase: Computational studies using real-world data demonstrated substantial potential revenue improvements. For instance, one customer's expected revenue increased from $5.75 (baseline) to $11.99 (optimized route) for promotional items not on their shopping list.
  • Time Reduction: Optimized routes using TSP strategies (like Lin-Kernighan) showed significant travel time reductions compared to baseline strategies, ranging from 5% to 11%.
  • Promotional Level Impact: Revenue increases with higher promotional levels, with a notable jump occurring when promotions increase from 3% to 6%.
  • Time Tolerance Sensitivity: Revenue increases with higher time tolerance factors, with the most significant jump observed between 5% and 10% tolerance.
  • Estimated Annual Revenue: Based on average customer traffic and a conservative strategy, an increased annual revenue of $500,000 was estimated for a single store.

Limitations and Future Work

  • RFID Cost and Adoption: The paper acknowledges that RFID technology, while powerful, is currently prohibitive in cost for widespread deployment at the scale discussed.
  • Simulated Promotions: Promotions were randomly generated, which might not fully capture the complexities of real-world promotion strategies.
  • Customer Behavior Nuances: While discrete choice models are used, the paper could further explore more sophisticated models to capture complex customer decision-making processes.
  • Future Work: The authors suggest that while RFID adoption might take longer, the concept of shopping lists deployable on PDAs or smartphones is likely to be adopted in the short term. They also point to the potential of combining smart shopping carts with displays for similar functionalities without full RFID implementation. Further research could explore more advanced routing algorithms and more nuanced customer behavior modeling.

Summary updated

BibTeX
@article{klabjan2010store,
  author = {Diego Klabjan and Jin-Xiang PEI},
  title = {{In-store one-to-one marketing}},
  journal = {Journal of Retailing and Consumer Services},
  year = {2010},
  doi = {10.1016/j.jretconser.2010.09.012},
}
Figure